Contents

Microsoft Security Best Practices

Introduction
What's inside Microsoft Security Best Practices?
Progress and role tracking worksheets
Governance, risk, and compliance
Videos and slides
Article
Capabilities
Security operations
Videos and slides
Article
Capabilities
Incident response
Overview
Planning
Process
Playbooks
Introduction
Phishing
Password spray
App consent grant
DART approach to ransomware
Identity and access management
Videos and slides
Article
Capabilities
Network security & containment
Videos and slides
Article
 Capabilities
Privileged administration
Securing privileged access
Privileged access strategy
Measuring success
Security levels
Privileged access accounts
Intermediaries
Interfaces
Privileged access devices
Enterprise access model
Privileged access implementation
Modernization plan
ESAE Retirement
Videos and slides
Privileged access administration
Ransomware and extortion
Human operated ransomware
Rapidly protect against ransomware and extortion
Overview
Phase 1. Prepare your recovery plan
Phase 2. Limit the scope of damage
Phase 3. Make it hard to get in
Backup and restore plan for ransomware
Information protection and storage
Videos and slides
Article
Capabilities
Applications and services
Article
 Microsoft Security Best Practices
12/13/2021 • 2 minutes to read • Edit Online

Microsoft Security Best Practices (formerly known as the Azure Security Compass or Microsoft Security
Compass) is a collection of best practices that provide clear actionable guidance for security related decisions.
This is designed to help you increase your security posture and reduce risk whether your environment is cloud-
only, or a hybrid enterprise spanning cloud(s) and on-premises data centers. This guidance was formerly
referred to as Azure Security Compass and is now increasing in scope to encompass all Microsoft security
guidance and capabilities, including Microsoft 365.
In this guidance:
Introduction
Governance, risk, and compliance
Security operations
Identity and access management
Network security and containment
Information protection and storage
Applications and services
This guidance is presented in a series of videos. To download the PowerPoint slides associated with these videos,
click download presentation slides.

Related topics
Security design principles for cloud architecture
Security reference architectures and design

Next steps
For additional security guidance from Microsoft, see Microsoft security documentation.
 What's inside Microsoft Security Best Practices?
12/13/2021 • 2 minutes to read • Edit Online

Microsoft Security Best Practices is a collection of best practices that provide clear actionable guidance for
security related decisions. This is designed to help you increase your security posture and reduce risk whether
your environment is cloud-only, or a hybrid enterprise spanning cloud(s) and on-premises data centers.
To download the PowerPoint slides associated with these videos, click download presentation slides.

Introduction to Microsoft Security Best Practices (14:58)

Evolution of Threat Environment, Roles, & Digital Strategies (20:04)

Transformation of Security, Strategies, Tools, & Threats (15:13)

Azure Regions and Services (02:59)

Microsoft Security Practices (13:49)

Azure Components and Reference Model (21:51)

Microsoft Security Best Practices workshop presentation

View the slides for this workshop.

PowerPoint | PDF

NOTE
Azure Security Compass or Microsoft Security Compass is now renamed as Microsoft Security Best Practices.

Top 10 Azure Security best practices

View slides for Top 10 Azure Security best practices.

PowerPoint | PDF
Next steps
For additional security guidance from Microsoft, see Microsoft security documentation.
 Progress and role tracking worksheets
12/13/2021 • 2 minutes to read • Edit Online

This topic includes tracking worksheets that assist you with tracking the status of decisions and roles.
These can be used by an organization to track status for Microsoft’s recommendations.

Tracking worksheet
The tracking worksheet assists with tracking status of decisions, policy, and implementation.

Excel

Cloud role tracking worksheet

The cloud role tracking worksheet assists with designating the parties responsible for specific functions in Azure.

PDF | PowerPoint

Next steps
For additional security guidance from Microsoft, see Microsoft security documentation.
 Microsoft Security Best Practices module:
Governance, risk, and compliance
12/13/2021 • 2 minutes to read • Edit Online

Governance, Risk, and Compliance (GRC) activities help reduce organizational risk by ensuring policy and best
practices are followed consistently over time. This section also addresses key roles and responsibilities we have
found important for successfully managing cloud security.
See the Governance, risk, and compliance and Capabilities topics for more information.
The following videos provide guidance on governance, risk, and compliance. You can also download the
PowerPoint slides associated with these videos.

Part 1: Introduction + Manage Connected Tenants (08:45)

Part 2: Clear Lines of Responsibility (02:46)

Part 3: Segmentation Strategy (02:11)

Part 4: Management Groups (04:15)

Part 5: Root Management Group (03:06)

Part 6: GRC Top Risks (03:31)

Part 7: Security Incident Notification (03:35)

Part 8: Access Reviews (02:15)

Part 9: Security Posture Improvement (03:30)

Part 10: Access for Security Personnel (03:18)

Part 11: Insecure Legacy Protocols (01:53)

Part 12: Compliance (04:29)

Part 13: Benchmarks (01:37)

Part 14: Azure Policy (02:30)

Part 15: Elevated Security Capabilities (03:43)

Part 16: General Guidance (03:01)

Next steps
For additional security guidance from Microsoft, see Microsoft security documentation.
 Governance, risk, and compliance
12/13/2021 • 20 minutes to read • Edit Online

Organizations of all sizes are constrained by their available resources; financial, people, and time. To achieve an
effective return on investment (ROI) organizations must prioritize where they will invest. Implementation of
security across the organization is also constrained by this, so to achieve an appropriate ROI on security the
organization needs to first understand and define its security priorities.
Governance – How is the organization’s security going to be monitored, audited, and reported? Design and
implementation of security controls within an organization is only the beginning of the story. How does the
organization know that things are actually working? Are they improving? Are there new requirements? Is there
mandatory reporting? Similar to compliance there may be external industry, government or regulatory
standards that need to be considered.
Risk – What types of risks does the organization face while trying to protect identifiable information, Intellectual
Property (IP), financial information? Who may be interested or could leverage this information if stolen,
including external and internal threats as well as unintentional or malicious? A commonly forgotten but
extremely important consideration within risk is addressing Disaster Recovery and Business Continuity.
Compliance – Are there specific industry, government, or regulatory requirements that dictate or provide
recommendation on criteria that your organization’s security controls must meet? Examples of such standards,
organizations, controls, and legislation are ISO27001, NIST, PCI-DSS.
The collective role of organization(s) is to manage the security standards of the organization through their
lifecycle:
Define - Set organizational standards and policies for practices, technologies, and configurations based
on internal factors (organizational culture, risk appetite, asset valuation, business initiatives, etc.) and
external factors (benchmarks, regulatory standards, threat environment, and more)
Improve – Continually push these standards incrementally forward towards the ideal state to ensure
continual risk reduction.
Sustain – Ensure the security posture doesn’t degrade naturally over time by instituting auditing and
monitoring compliance with organizational standards.

Prioritize security best practices investments

Security best practices are ideally applied proactively and completely to all systems as you build your cloud
program, but this isn’t reality for most enterprise organizations. Business goals, project constraints, and other
factors often cause organizations to balance security risk against other risks and apply a subset of best practices
at any given point.
We recommend applying as many as of the best practices as early as possible, and then working to retrofit any
gaps over time as you mature your security program. We recommend evaluating the following considerations
when prioritizing which to follow first:
High business impact and highly exposed systems – These include systems with direct intrinsic
value as well as the systems that provide attackers a path to them. For more information, see Identify and
classify business critical applications.
Easiest to implement Mitigations – Identify quick wins by prioritizing the best practices, which your
organization can execute quickly because you already have the required skills, tools, and knowledge to do
 it (for example, implementing a Web App Firewall (WAF) to protect a legacy application).
Be careful not to exclusively use (or overuse) this short-term prioritization method. Doing so can increase
your risk by preventing your program from growing and leaving critical risks exposed for extended
periods.
Microsoft has provided some prioritized lists of security initiatives to help organizations start with these
decisions based on our experience with threats and mitigation initiatives in our own environments and across
our customers. See Module 4a of the Microsoft CISO Workshop

Manage connected tenants

Ensure your security organization is aware of all enrollments and associated subscriptions connected to your
existing environment (via ExpressRoute or Site-Site VPN) and monitoring as part of the overall enterprise.
These azure resources are part of your enterprise environment and security organizations require visibility into
them. Security organizations need this access to assess risk and to identify whether organizational policies and
applicable regulatory requirements are being followed.
Ensure all Azure environments that connect to your production environment/network apply your organization’s
policy and IT governance controls for security. You can discover existing connected tenants using a tool provided
by Microsoft. Guidance on permissions you may assign to security is in the Assign privileges for managing the
environment section.

Clear lines of responsibility

Designate the parties responsible for specific functions in Azure
Clearly documenting and sharing the contacts responsible for each of these functions will create consistency
and facilitate communication. Based on our experience with many cloud adoption projects, this will avoid
confusion that can lead to human and automation errors that create security risk.
Designate groups (or individual roles) that will be responsible for these key functions:

GRO UP O R IN DIVIDUA L RO L E RESP O N SIB IL IT Y

Network Security Typically existing network security team. Configuration and

maintenance of Azure Firewall, Network Virtual Appliances
(and associated routing), WAFs, NSGs, ASGs, etc.

Network Management Typically existing network operations team. Enterprise-wide

virtual network and subnet allocation.

Ser ver Endpoint Security Typically IT operations, security, or jointly. Monitor and
remediate server security (patching, configuration, endpoint
security, etc.).

Incident Monitoring and Response Typically security operations team. Investigate and remediate
security incidents in Security Information and Event
Management (SIEM) or source console.

Policy Management Typically GRC team + Architecture. Set Direction for use of
Role Based Access Control (RBAC), Microsoft Defender for
Cloud, Administrator protection strategy, and Azure Policy to
govern Azure resources.
 GRO UP O R IN DIVIDUA L RO L E RESP O N SIB IL IT Y

Identity Security and Standards Typically Security Team + Identity Team jointly. Set direction
for Azure AD directories, PIM/PAM usage, MFA,
password/synchronization configuration, Application Identity
Standards.

Enterprise segmentation strategy

Identify groups of resources that can be isolated from other parts of the enterprise to contain (and detect)
adversary movement within your enterprise. This unified enterprise segmentation strategy will guide all
technical teams to consistently segment access using networking, applications, identity, and any other access
controls.
A clear and simple segmentation strategy helps contain risk while enabling productivity and business
operations.
An enterprise segmentation strategy is defined higher than a traditional “network segmentation” security
strategy. Traditional segmentation approaches for on premises environments frequently failed to achieve their
goals because they were developed “bottom-up” by different technical teams and were not aligned well with
business use cases and application workloads. This resulted in overwhelming complexity that generates support
issues and often undermines the original purpose with broad network firewall exceptions.
Creating a unified enterprise segmentation strategy enables to guide all technical teams stakeholders (IT,
Security, Applications, etc.) Business Units that is built around the business risks and needs will increase
alignment to and understand and support sustainability of the security containment promises. This clarity and
alignment will also reduce s the risk of human errors and automation failures that can lead to security
vulnerabilities, operational downtime, or both.
While network micro-segmentation also offers promise to reduce risk (discussed more in Network Security and
Containment section), it doesn’t eliminate the need to align technical teams. Micro segmentation should be
considered after to and plans to ensure the ensuring technical teams are aligned so you can avoid a recurrence
of the internal conflicts that plagued and confusion of the on-premises network generation segmentation
strategies.
Here are Microsoft's recommendations for prioritizing initiatives on containment and segmentation (based on
Zero Trust principles). These recommendations are listed in priority order by highest importance.
Ensure alignment of technical teams to a single enterprise segmentation strategy.
Invest in broadening containment by establishing a modern perimeter based on zero trust principles
focused on identity, device, applications, and other signals (to overcome limitation of network controls to
protect new resources and attack types).
Bolster network controls for legacy applications by exploring micro segmentation strategies.
A good enterprise segmentation strategy meets these criteria:
Enables Operations – Minimizes operation friction by aligning to business practices and applications
Contains Risk - Adds cost and friction to attackers by
Isolating sensitive workloads from compromise of other assets
Isolating high exposure systems from being used as a pivot to other systems
Monitored – Security Operations should monitor for potential violations of the integrity of the segments
(account usage, unexpected traffic, etc.)
Security team visibility
Provide security teams read-only access to the security aspects of all technical resources in their purview
Security organizations require visibility into the technical environment to perform their duties of assessing and
reporting on organizational risk. Without this visibility, security will have to rely on information provided from
groups operating the environment who have a potential conflict of interest (and other priorities).
Note that security teams may separately be granted additional privileges if they have operational
responsibilities or a requirement to enforce compliance on Azure resources.
For example in Azure, assign security teams to the Security Readers permission that provides access to
measure security risk (without providing access to the data itself)
For enterprise security groups with broad responsibility for security of Azure, you can assign this permission
using:
Root management group – for teams responsible for assessing and reporting risk on all resources
Segment management group(s) – for teams with limited scope of responsibility (typically required
because of organizational boundaries or regulatory requirements)
Because security will have broad access to the environment (and visibility into potentially exploitable
vulnerabilities), you should consider them critical impact accounts and apply the same protections as
administrators. The Administration section details these controls for Azure.

Assign privileges for managing the environment

Grant roles with operational responsibilities in Azure the appropriate permissions based on a clearly
documented strategy built from the principle of least privilege and your operational needs.
Providing clear guidance that follows a reference model will reduce risk because by increasing it provides clarity
for your technical teams implementing these permissions. This clarity makes it easier to detect and correct
human errors like overpermissioning, reducing your overall risk.
Microsoft recommends starting from these Microsoft reference models and adapting to your organization.
Core Services Reference Permissions
This segment hosts shared services utilized across the organization. These shared services typically include
Active Directory Domain Services, DNS/DHCP, System Management Tools hosted on Azure Infrastructure as a
Service (IaaS) virtual machines.
Security Visibility across all resources – For security teams, grant read-only access to security attributes for
all technical environments. This access level is needed to assess risk factors, identify potential mitigations, and
advise organizational stakeholders who accept the risk. See Security Team Visibility for more details.
Policy management across some or all resources – To monitor and enforce compliance with external (or
internal) regulations, standards, and security policy, assign appropriate permission to those roles. The roles and
permissions you choose will depend on the organizational culture and expectations of the policy program. See
Microsoft Cloud Adoption Framework for Azure.
Central IT operations across all resources – Grant permissions to the central IT department (often the
infrastructure team) to create, modify, and delete resources like virtual machines and storage.
Central networking group across network resources – To ensure consistency and avoid technical conflicts,
assign network resource responsibilities to a single central networking organization. These resources should
include virtual networks, subnets, Network Security Groups (NSG), and the virtual machines hosting virtual
network appliances. See Centralize Network Management And Security for more details
Resource Role Permissions – For most core services, administrative privileges required to manage them are
granted via the application itself (Active Directory, DNS/DHCP, System Management Tools, etc.), so no additional
Azure resource permissions are required. If your organizational model requires these teams to manage their
own VMs, storage, or other Azure resources, you can assign these permissions to those roles.
Ser vice admin (Break Glass Account) – Use the service admin role only for emergencies (and initial setup if
required). Do not use this role for daily tasks. See Emergency Access (‘Break Glass’ Accounts) for more details.
Segment reference permissions
This segment permission design provides consistency while allowing flexibility to accommodate the range of
organizational models from a single centralized IT group to mostly independent IT and DevOps teams.
Security visibility across all resources – For security teams, grant read-only access to security attributes for
all technical environments. This access level is needed to assess risk factors, identify potential mitigations, and
advise organizational stakeholders who accept the risk. See Security Team Visibility.
Policy management across some or all resources – To monitor and enforce compliance with external (or
internal) regulations, standards, and security policy assign appropriate permission to those roles. The roles and
permissions you choose will depend on the organizational culture and expectations of the policy program. See
Microsoft Cloud Adoption Framework for Azure.
IT Operations across all resources – Grant permission to create, modify, and delete resources. The purpose
of the segment (and resulting permissions) will depend on your organization structure.
Segments with resources managed by a centralized IT organization can grant the central IT department
(often the infrastructure team) permission to modify these resources.
Segments managed by independent business units or functions (such as a Human Resources IT Team)
can grant those teams permission to all resources in the segment.
Segments with autonomous DevOps teams don’t need to grant permissions across all resources because
the resource role (below) grants permissions to application teams. For emergencies, use the service
admin account (break-glass account).
Central networking group across network resources – To ensure consistency and avoid technical conflicts,
assign network resource responsibilities to a single central networking organization. These resources should
include virtual networks, subnets, Network Security Groups (NSG), and the virtual machines hosting virtual
network appliances. See Centralize Network Management And Security.
Resource Role Permissions – Segments with autonomous DevOps teams will manage the resources
associated with each application. The actual roles and their permissions depend on the application size and
complexity, the application team size and complexity, and the culture of the organization and application team.
Ser vice Admin (Break Glass Account) – Use the service admin role only for emergencies (and initial setup if
required). Do not use this role for daily tasks. See Emergency Access (‘Break Glass’ Accounts) for more details.
Permission Guidance and Tips
To drive consistency and ensure application to future subscriptions, permissions should be assigned at
management group for the segment rather than the individual subscriptions. See Avoid Granular and
Custom Permissions for more details.
You should first review the built-in roles to see if one is applicable before creating a custom role to grant
the appropriate permissions to VMs and other objects. See Use Built in Roles for more details
Security managers group membership may be appropriate for smaller teams/organizations where
security teams have extensive operational responsibilities.

Establish segmentation with management groups

Structure management groups into a simple design that guides the enterprise segmentation model.
Management groups offer the ability to consistently and efficiently manage resources (including multiple
subscriptions as needed). However, because of their flexibility, it's possible to create an overly complex design.
Complexity creates confusion and negatively impacts both operations and security (as illustrated by overly
complex Organizational Unit (OU) and Group Policy Object (GPO) designs for Active Directory).
Microsoft recommends aligning the top level of management groups (MGs) into a simple enterprise
segmentation strategy limited to 1 or 2 levels.

Use root management group carefully

Use the Root Management Group (MG) for enterprise consistency, but test changes carefully to minimize risk of
operational disruption.
The root management group enables you to ensure consistency across the enterprise by applying policies,
permissions, and tags across all subscriptions. Care must be taken when planning and implementing
assignments to the root management group because this can affect every resource on Azure and potentially
cause downtime or other negative impacts on productivity in the event of errors or unanticipated effects.
Root management group guidance:
Plan Carefully - Select enterprise-wide elements to the root management group that have a clear
requirement to be applied across every resource and/or low impact.
Good candidates include:
Regulator y requirements with clear business risk/impact (for example, restrictions related to
data sovereignty).
Near-zero potential negative impact on operations such as policy with audit effect, Tag
assignment, RBAC permissions assignments that have been carefully reviewed.
Test First - Carefully test all enterprise-wide changes on the root management group before applying
(policy, tags, RBAC model, etc.) using a
Test Lab - Representative lab tenant or lab segment in production tenant.
Production Pilot - Segment MG or Designated subset in subscription(s) / MG.
Validate Changes – to ensure they have the desired effect.

Virtual Machine (VM) security updates and strong passwords

Ensure policy and processes enable (and require) rapid application of security updates to virtual machines.
Attackers constantly scan public cloud IP ranges for open management ports and attempt “easy” attacks like
common passwords and unpatched vulnerabilities.
Enable Microsoft Defender for Cloud to identify missing security updates & apply them.
Local Admin Password Solution (LAPS) or a third party Privileged Access Management can set strong local
admin passwords and just in time access to them.

Remove Virtual Machine (VM) direct internet connectivity

Ensure policy and processes require restricting and monitoring direct internet connectivity by virtual machines
Attackers constantly scan public cloud IP ranges for open management ports and attempt “easy” attacks like
common passwords and known unpatched vulnerabilities
This can be accomplished with one or more methods in Azure:
Enterprise-wide prevention - Prevent inadvertent exposure with an enterprise network and
permission model such as the reference model described throughout this guidance. This significantly
reduces the risk of accidental VM internet exposure by
Ensuring that network traffic is routed through approved egress points by default
Exceptions (for example, add a public IP address to a resource) must go through a centralized
group (which can carefully evaluate exception requests to ensure appropriate controls are applied)
Identify and Remediate exposed VMs using the Microsoft Defender for Cloud network visualization to
quickly identify internet exposed resources.
Restrict management por ts (RDP, SSH) using Just in Time access in Microsoft Defender for Cloud.

Assign incident notification contact

Ensure a security contact receives Azure incident notifications from Microsoft typically a notification that your
resource is compromised and/or attacking another customer.
This enables your security operations team to rapidly respond to potential security risks and remediate them.
Ensure administrator contact information in the Azure enrollment portal includes contact information that will
notify security operations (directly or rapidly via an internal process)

Regularly review critical access

Regularly review roles that are assigned privileges with a business-critical impact.
Set up a recurring review pattern to ensure that accounts are removed from permissions as roles change. You
can conduct the review manually or through an automated process by using tools such as Azure AD access
reviews.

Discover and remediate common risks

Identity well known risks for your Azure tenants, remediate those risks, and track your progress using Secure
Score.
Identifying and remediating common security hygiene risks significantly reduces overall risk to your
organization by increasing cost to attackers. When you remove cheap and well-established attack vectors,
attackers are forced to acquire and use advanced or untested attack methods.
Azure Secure Score in Microsoft Defender for Cloud monitors the security posture of machines, networks,
storage and data services, and applications to discover potential security issues (internet connected VMs, or
missing security updates, missing endpoint protection or encryption, deviations from baseline security
configurations, missing Web Application Firewall (WAF), and more). You should enable this capability (no
additional cost), review the findings, and follow the included recommendations to plan and execute technical
remediations starting with the highest priority items.
As you address risks, track progress and prioritize ongoing investments in your governance and risk reduction
programs.

Increase automation with Azure Blueprints

Use Azure’s native automation capabilities to increase consistency, compliance, and deployment speed for
workloads.
Automation of deployment and maintenance tasks reduces security and compliance risk by limiting opportunity
to introduce human errors during manual tasks. This will also allow both IT Operations teams and security
teams to shift their focus from repeated manual tasks to higher value tasks like enabling developers and
business initiatives, protecting information, and so on.
Utilize the Azure Blueprint service to rapidly and consistently deploy application environments that are
compliant with your organization’s policies and external regulations. Azure Blueprint Service automates
deployment of environments including RBAC roles, policies, resources (VM/Net/Storage/etc.), and more. Azure
Blueprints builds on Microsoft’s significant investment into the Azure Resource Manager to standardize resource
deployment in Azure and enable resource deployment and governance based on a desired-state approach. You
can use built in configurations in Azure Blueprint, make your own, or just use Resource Manager scripts for
smaller scope.
Several Security and Compliance Blueprints samples are available to use as a starting template.

Evaluate security using benchmarks

Use an industry standard benchmark to evaluate your organizations current security posture.
Benchmarking allows you to improve your security program by learning from external organizations.
Benchmarking lets you know how your current security state compares to that of other organizations, providing
both external validation for successful elements of your current system as well as identifying gaps that serve as
opportunities to enrich your team’s overall security strategy. Even if your security program isn’t tied to a specific
benchmark or regulatory standard, you will benefit from understanding the documented ideal states by those
outside and inside of your industry.
As an example, the Center for Internet Security (CIS) has created security benchmarks for Azure that map to
the CIS Control Framework. Another reference example is the MITRE ATT&CK™ framework that defines the
various adversary tactics and techniques based on real-world observations. These external references control
mappings help you to understand any gaps between your current strategy what you have and what other
experts in the industry.

Audit and enforce policy compliance

Ensure that the security team is auditing the environment to report on compliance with the security policy of the
organization. Security teams may also enforce compliance with these policies.
Organizations of all sizes will have security compliance requirements. Industry, government, and internal
corporate security policies all need to be audited and enforced. Policy monitoring is critical to check that initial
configurations are correct and that it continues to be compliant over time.
In Azure, you can take advantage of Azure Policy to create and manage policies that enforce compliance. Like
Azure Blueprints, Azure Policies are built on the underlying Azure Resource Manager capabilities in the Azure
platform (and Azure Policy can also be assigned via Azure Blueprints).
For more information on how to do this in Azure, please review Tutorial: Create and manage policies to
enforce compliance .

Monitor identity Risk

Monitor identity related risk events for warning on potentially compromised identities and remediate those
risks.
Most security incidents take place after an attacker initially gains access using a stolen identity. These identities
can often start with low privileges, but the attackers then use that identity to traverse laterally and gain access to
more privileged identities. This repeats as needed until the attacker controls access to the ultimate target data or
systems.
Azure Active Directory uses adaptive machine learning algorithms, heuristics, and known compromised
credentials (username/password pairs) to detect suspicious actions that are related to your user accounts. These
username/password pairs come from monitoring public and dark web sites (where attackers often dump
compromised passwords) and by working with security researchers, law enforcement, Security teams at
Microsoft, and others.
There are two places where you review reported risk events:
Azure AD repor ting - Risk events are part of Azure AD's security reports. For more information, see the
users at risk security report and the risky sign-ins security report.
Azure AD Identity Protection - Risk events are also part of the reporting capabilities of Azure Active
Directory Identity Protection.
In addition, you can use the Identity Protection risk events API to gain programmatic access to security
detections using Microsoft Graph.
Remediate these risks by manually addressing each reported account or by setting up a user risk policy to
require a password change for these high risk events.

Penetration testing
Use Penetration Testing to validate security defenses.
Real world validation of security defenses is critical to validate your defense strategy and implementation. This
can be accomplished by a penetration test (simulates a one time attack) or a red team program (simulates a
persistent threat actor targeting your environment).
Follow the guidance published by Microsoft for planning and executing simulated attacks.

Discover & replace insecure protocols

Discover and disable the use of legacy insecure protocols SMBv1, LM/NTLMv1, wDigest, Unsigned LDAP Binds,
and Weak ciphers in Kerberos.
Authentication protocols are a critical foundation of nearly all security assurances. These older versions can be
exploited by attackers with access to your network and are often used extensively on legacy systems on
Infrastructure as a Service (IaaS).
Here are ways to reduce your risk:
Discover protocol usage by reviewing logs with Microsoft Sentinel’s Insecure Protocol Dashboard or
 third party tools
Restrict or Disable use of these protocols by following guidance for SMB, NTLM, WDigest
We recommend implementing changes using pilot or other testing method to mitigate risk of operational
interruption.

Elevated security capabilities

Consider whether to utilize specialized security capabilities in your enterprise architecture.
These measures have the potential to enhance security and meet regulatory requirements, but can introduce
complexity that may negatively impact your operations and efficiency.
We recommend careful consideration and judicious use of these security measures as required:
Dedicated Hardware Security Modules (HSMs)
Dedicated Hardware Security Modules (HSMs) may help meet regulatory or security requirements.
Confidential Computing
Confidential Computing may help meet regulatory or security requirements.

Next steps
For additional security guidance from Microsoft, see Microsoft security documentation.
 Governance, risk, and compliance capabilities
12/13/2021 • 2 minutes to read • Edit Online

This article lists capabilities that can help with governance, risk, and compliance. You can also learn more about
these capabilities at Azure governance documentation.

C A PA B IL IT Y DESC RIP T IO N M O RE IN F O RM AT IO N

Microsoft 365 Defender portal Microsoft 365 Defender portal Microsoft 365 Defender portal
provides security administrators and
other risk management professionals
with a centralized hub and specialized
workspace that enables them to
manage and take full advantage of
Microsoft 365 intelligent security
solutions for identity and access
management, threat protection,
information protection, and security
management.

Microsoft 365 compliance center Microsoft 365 compliance center Microsoft 365 compliance center
provides easy access to the data and
tools you need to manage to your
organization's compliance needs.

Microsoft Defender for Cloud Microsoft Defender for Cloud is a Microsoft Defender for Cloud
unified infrastructure security documentation
management system that strengthens
the security posture of your data
centers, and provides advanced threat
protection across your hybrid
workloads in the cloud - whether
they're in Azure or not - as well as on
premises.

Management Groups If your organization has many Organize your resources with Azure
subscriptions, you may need a way to management groups
efficiently manage access, policies, and
compliance for those subscriptions.
Azure management groups provide a
level of scope above subscriptions. You
organize subscriptions into containers
called "management groups" and
apply your governance conditions to
the management groups.

Azure Policy Azure Policy is a service in Azure that Azure Policy documentation
you use to create, assign, and manage
policies. These policies enforce different
rules and effects over your resources,
so those resources stay compliant with
your corporate standards and service
level agreements.
 C A PA B IL IT Y DESC RIP T IO N M O RE IN F O RM AT IO N

Azure Blueprints Azure Blueprints enables cloud Azure Blueprints documentation

architects and central information
technology groups to define a
repeatable set of Azure resources that
implements and adheres to an
organization's standards, patterns, and
requirements. Azure Blueprints makes
it possible for development teams to
rapidly build and stand up new
environments with trust they're
building within organizational
compliance with a set of built-in
components -- such as networking --
to speed up development and delivery.

Next steps
For additional security guidance from Microsoft, see Microsoft security documentation.
 Security operations
12/13/2021 • 2 minutes to read • Edit Online

Security operations monitor an enterprise environment to rapidly identify and remediate risk from active attack
operations, sharing insights and threat intelligence from these attacks to the rest of the organization.
The following videos provide guidance on security operations.

Part 1: Introduction - SOC Learnings, Strategies, and Technical

Integration (24:30 long)

Part 2: Azure Alerts (2:36 long)

Part 3: Alert and Log Ingestion (4:51 long)

Part 4: Journey to Cloud Analytics (6:05 long)

Part 5: Security Operations General Guidance (3:42 long)

Next steps
See security operations best practices and capabilities for more information.
Review the PowerPoint slides for the Microsoft Azure Security Compass Workshop.

See also
Security operations functions from the Cloud Adoption Framework for Azure
SOC Process Framework Workbook for Microsoft Sentinel
Additional security guidance from Microsoft

Key Microsoft security resources

RESO URC E DESC RIP T IO N

2021 Microsoft Digital Defense Report A report that encompasses learnings from security experts,
practitioners, and defenders at Microsoft to empower people
everywhere to defend against cyberthreats.
RESO URC E DESC RIP T IO N

Microsoft Cybersecurity Reference Architectures A set of visual architecture diagrams that show Microsoft’s
cybersecurity capabilities and their integration with Microsoft
cloud platforms such as Microsoft 365 and Microsoft Azure
and third-party cloud platforms and apps.

Minutes matter infographic download An overview of how Microsoft's SecOps team does incident
response to mitigate ongoing attacks.

Azure Cloud Adoption Framework security operations Strategic guidance for leaders establishing or modernizing a
security operation function.

Microsoft cloud security for IT architects model Security across Microsoft cloud services and platforms for
identity and device access, threat protection, and
information protection.

Microsoft security documentation Additional security guidance from Microsoft.

 Microsoft security best practices for security
operations
12/13/2021 • 7 minutes to read • Edit Online

Security operations maintain and restore the security assurances of the system as live adversaries attack it. The
tasks of security operations are described well by the NIST Cybersecurity Framework functions of Detect,
Respond, and Recover.
Detect - Security operations must detect the presence of adversaries in the system, who are incentivized
to stay hidden in most cases as this allows them to achieve their objectives unimpeded. This can take the
form of reacting to an alert of suspicious activity or proactively hunting for anomalous events in the
enterprise activity logs.
Respond – Upon detection of potential adversary action or campaign, security operations must rapidly
investigate to identify whether it is an actual attack (true positive) or a false alarm (false positive) and
then enumerate the scope and goal of the adversary operation.
Recover – The ultimate goal of security operations is to preserve or restore the security assurances
(confidentiality, integrity, availability) of business services during and after an attack.
The most significant security risk most organizations face is from human attack operators (of varying skill
levels). This is because risk from automated/repeated attacks have been mitigated significantly for most
organizations by signature and machine learning based approaches built into anti-malware (though there are
notable exceptions like Wannacrypt and NotPetya, which moved faster than these defenses).
While human attack operators are challenging to face because of their adaptability (vs. automated/repeated
logic), they are operating at the same “human speed” as defenders, which help level the playing field.
Security Operations (sometimes referred to as a Security Operations Center (SOC)) has a critical role to play in
limiting the time and access an attacker can get to valuable systems and data. Each minute that an attacker has
in the environment allows them to continue to conduct attack operations and access sensitive or valuable
systems.

Objective and metrics

The metrics you measure will have a significant effect on the behaviors and outcomes of security operations.
Focusing on the right measurements will help drive continuous improvement in the right areas that
meaningfully reduce risk.
To ensure that security operations are effectively containing attackers access, the objectives should focus on:
Reducing time to acknowledge an alert to ensure that detected adversaries are not ignored while
defenders are spending time investigating false positives.
Reducing time to remediate a discovered adversary to reduce their opportunity time to conduct and
attack and reach sensitive systems
Prioritizing security investments into systems that have high intrinsic value (likely targets or high
business impact) and access to many systems or sensitive systems (administrator accounts and sensitive
users)
Increasing focus on proactively hunting for adversaries as your program matures and reactive
incidents get under control. This is focused on reducing the time that a higher skilled adversary can
 operate in the environment (for example, skilled enough to evade reactive alerts).
For more information on how Microsoft’s SOC uses these metrics, see https://aka.ms/ITSOC.

Hybrid enterprise view

Security operations should ensure their tooling, processes, and analyst skill sets enable visibility across the full
span of their hybrid estate.
Attackers don’t restrict their actions to a particular environment when targeting an organization, they will attack
resources on any platform using any method available. Enterprise organizations adopting cloud services like
Azure and AWS are effectively operating a hybrid of cloud and on-premises assets.
Security operations tooling and processes should be designed for attacks on cloud and on-premises assets as
well as attackers pivoting between cloud and on-premises resources using identity or other means. This
enterprise-wide view will enable security operations teams to rapidly detect, respond, and recover from attacks,
reducing organizational risk.

Leverage native detections and controls

You should favor the use of security detections and controls built into the cloud platform before creating custom
detections using event logs from the cloud.
Cloud platforms evolve rapidly with new features, which can make maintaining detections challenging. Native
controls are maintained by the cloud provider and are typically high quality (low false positive rate).
Because many organizations may use multiple cloud platforms and need a unified view across the enterprise,
you should ensure these native detections and controls feed a centralized SIEM or other tool. We don’t
recommend trying to substitute generalized log analysis tools and queries instead of native detections and
controls. These tools can offer numerous values for proactive hunting activities but getting to a high-quality
alert with these tools requires application of deep expertise and time that could be better spent on hunting and
other activities.
To complement the broad visibility of a centralized SIEM (such as Microsoft Sentinel, Splunk, or QRadar), you
should leverage native detections and controls such as:
Organizations using Azure should leverage capabilities like Microsoft Defender for Cloud for alert
generation on the Azure platform.
Organizations should leverage native logging capabilities like Azure Monitor and AWS CloudTrail for
pulling logs into a central view.
Organizations using Azure should leverage Network Security Group (NSG) capabilities for visibility into
network activities on the Azure platform.
Investigation practices should leverage native tools with deep knowledge of the asset type such as an
Endpoint Detection and Response (EDR) solution, identity tools, and Microsoft Sentinel.

Prioritize alert and log integration

Ensure that you are integrating critical security alerts and logs into SIEMs without introducing a high volume of
low value data.
Introducing too much low-value data can increase SIEM cost, increase noise and false positives, and lower
performance.
The data you collect should be focused on supporting one or more of these operations activities:
 Aler ts (detections from existing tools or data required for generating custom alerts)
Investigation of an incident (for example, required for common queries)
Proactive hunting activities
Integrating more data can allow you to enrich alerts with additional context that enable rapid response and
remediation (filter false positives, and elevate true positives, etc.), but collection is not detection. If you don’t
have a reasonable expectation that the data will provide value (for example, high volume of firewall denies
events), you may deprioritize integration of these events.

SecOps resources for Microsoft security services

If you are new-to-role as a security analyst, see these resources to get you started.

TO P IC RESO URC E

SecOps planning for incident response Incident response planning for preparing your organization
for an incident.

SecOps incident response process Incident response process for best practices on responding
to an incident.

Incident response workflow Example incident response workflow for Microsoft 365
Defender

Periodic security operations Example periodic security operations for Microsoft 365
Defender

Investigation for Microsoft Sentinel Incidents in Microsoft Sentinel

Investigation for Microsoft 365 Defender Incidents in Microsoft 365 Defender

If you are an experienced security analyst, see these resources to quickly ramp up your SecOps team for
Microsoft security services.

TO P IC RESO URC E

Azure Active Directory (Azure AD) Security operations guide

Microsoft 365 Defender Security operations guide

Microsoft Sentinel How to investigate incidents

Microsoft 365 Defender How to investigate incidents

Security operations establishment or modernization Azure Cloud Adoption Framework articles for SecOps and
SecOps functions

Incident response playbooks Overview at https://aka.ms/IRplaybooks

- Phishing
- Password spray
- App consent grant
 TO P IC RESO URC E

SOC Process Framework Microsoft Sentinel

MSTICPy and Jupyter Notebooks Microsoft Sentinel

Blog series about SecOps within Microsoft

See this blog series about how the SecOps team at Microsoft works.
Part 1 – Organization: Mission and Culture
Part 2a – People: Teams, Tiers, and Roles
Part 2b – People: Careers and Readiness
Part 3a – Technology: SOC Tooling
Part 3b – Technology: Day in life of an analyst
Part 3c – A day in the life part 2 - Microsoft Security
Part 3d – Zen and the art of threat hunting
Simuland
Simuland is an open-source initiative to deploy lab environments and end-to-end simulations that:
Reproduce well-known techniques used in real attack scenarios.
Actively test and verify the effectiveness of related Microsoft 365 Defender, Microsoft Defender for Cloud,
and Microsoft Sentinel detections.
Extend threat research using telemetry and forensic artifacts generated after each simulation exercise.
Simuland lab environments provide use cases from a variety of data sources including telemetry from Microsoft
365 Defender security products, Microsoft Defender for Cloud, and other integrated data sources through
Microsoft Sentinel data connectors.
In the safety of a trial or paid sandbox subscription, you can:
Understand the underlying behavior and functionality of adversary tradecraft.
Identify mitigations and attacker paths by documenting preconditions for each attacker action.
Expedite the design and deployment of threat research lab environments.
Stay up to date with the latest techniques and tools used by real threat actors.
Identify, document, and share relevant data sources to model and detect adversary actions.
Validate and tune detection capabilities.
The learnings from Simuland lab environment scenarios can then be implemented in production.
After reading an overview of Simuland, see the Simuland GitHub repository.

Key Microsoft security resources

RESO URC E DESC RIP T IO N

2021 Microsoft Digital Defense Report A report that encompasses learnings from security experts,
practitioners, and defenders at Microsoft to empower people
everywhere to defend against cyberthreats.
 RESO URC E DESC RIP T IO N

Microsoft Cybersecurity Reference Architectures A set of visual architecture diagrams that show Microsoft’s
cybersecurity capabilities and their integration with Microsoft
cloud platforms such as Microsoft 365 and Microsoft Azure
and third-party cloud platforms and apps.

Minutes matter infographic download An overview of how Microsoft's SecOps team does incident
response to mitigate ongoing attacks.

Azure Cloud Adoption Framework security operations Strategic guidance for leaders establishing or modernizing a
security operation function.

Microsoft cloud security for IT architects model Security across Microsoft cloud services and platforms for
identity and device access, threat protection, and
information protection.

Microsoft security documentation Additional security guidance from Microsoft.

Next step
Review security operations capabilities.
 Security operations capabilities
12/13/2021 • 2 minutes to read • Edit Online

These are the capabilities that you can use for your security operations.

C A PA B IL IT Y DESC RIP T IO N M O RE IN F O RM AT IO N

Microsoft Sentinel A scalable, cloud-native, security Microsoft Sentinel documentation

information event management (SIEM)
and security orchestration automated
response (SOAR) solution. Microsoft
Sentinel delivers intelligent security
analytics and threat intelligence across
the enterprise, providing a single
solution for alert detection, threat
visibility, proactive hunting, and threat
response.

Microsoft Defender for Cloud A unified infrastructure security Microsoft Defender for Cloud
management system that strengthens documentation
the security posture of your data
centers, and provides advanced threat
protection across your hybrid
workloads in the cloud - whether
they're in Azure or not - as well as-on
premises.

Azure Active Directory (Azure AD) Azure AD Identity Protection enables What is Azure Active Directory Identity
Identity Protection you to detect potential vulnerabilities Protection?
affecting your organization's identities
and configure automated remediation
policy to low, medium, and high sign-
in risk and user risk.

Microsoft Defender for Identity A cloud-based security solution that Defender for Identity documentation
leverages your on-premises Active
Directory Domain Services signals to
identify, detect, and investigate
advanced threats, compromised
identities, and malicious insider actions
directed at your organization.
Defender for Identity empowers SecOp
analysts and security professionals to
detect advanced attacks in hybrid
environments.

Microsoft Defender for Office 365 Safeguards your organization against Defender for Office 365
malicious threats posed by email documentation
messages, links (URLs), and
collaboration tools.
 C A PA B IL IT Y DESC RIP T IO N M O RE IN F O RM AT IO N

Microsoft Defender for Endpoint An endpoint protection platform Defender for Endpoint documentation
designed to help enterprise networks
prevent, detect, investigate, and
respond to advanced threats.

Microsoft Defender for Cloud Apps A cloud access security broker (CASB) Microsoft Defender for Cloud Apps
that operates on multiple clouds. It documentation
provides rich visibility, control over
data travel, and sophisticated analytics
to identify and combat cyberthreats
across all your cloud services.

Azure Monitor Maximizes the availability and Azure Monitor documentation

performance of your applications and
services by delivering a comprehensive
solution for collecting, analyzing, and
acting on telemetry from your cloud
and on-premises environments. It
helps you understand how your
applications are performing and
proactively identifies issues affecting
them and the resources they depend
on.

Microsoft 365 Defender portal Combines protection, detection, Microsoft 365 Defender portal
investigation, and response to email, documentation
collaboration, identity, and device
threats, in a central portal. It includes
information from Defender for Office
365, Defender for Endpoint, Defender
for Identity, and Microsoft Defender
for Cloud Apps for quick access to
information, simpler layouts, and
bringing related information together
for easier alert detection, threat
visibility, proactive hunting, and
incident response.

See also
Security operations functions from the Cloud Adoption Framework for Azure
SOC Process Framework Workbook for Microsoft Sentinel
Azure AD security operations guide
Microsoft 365 Defender security operations guide

Key Microsoft security resources

RESO URC E DESC RIP T IO N

2021 Microsoft Digital Defense Report A report that encompasses learnings from security experts,
practitioners, and defenders at Microsoft to empower people
everywhere to defend against cyberthreats.
RESO URC E DESC RIP T IO N

Microsoft Cybersecurity Reference Architectures A set of visual architecture diagrams that show Microsoft’s
cybersecurity capabilities and their integration with Microsoft
cloud platforms such as Microsoft 365 and Microsoft Azure
and third-party cloud platforms and apps.

Minutes matter infographic download An overview of how Microsoft's SecOps team does incident
response to mitigate ongoing attacks.

Azure Cloud Adoption Framework security operations Strategic guidance for leaders establishing or modernizing a
security operation function.

Microsoft cloud security for IT architects model Security across Microsoft cloud services and platforms for
identity and device access, threat protection, and
information protection.

Microsoft security documentation Additional security guidance from Microsoft.

 Incident response overview
12/13/2021 • 3 minutes to read • Edit Online

Incident response is the practice of investigating and remediating active attack campaigns on your organization.
This is part of the security operations discipline and is primarily reactive in nature.
Incident response has the largest direct influence on the overall mean time to acknowledge (MTTA) and mean
time to remediate (MTTR) that measure how well security operations are able to reduce organizational risk.
Incident response teams heavily rely on good working relationships between threat hunting, intelligence, and
incident management teams (if present) to actually reduce risk. See SecOps metrics for more information.
For more information on security operations roles and responsibilities, see Cloud SOC functions.

New-to-role resources
If you are new-to-role as a security analyst, see these resources to get you started.

TO P IC RESO URC E

SecOps planning for incident response Incident response planning for preparing your organization
for an incident.

SecOps incident response process Incident response process for best practices on responding
to an incident.

Incident response workflow Example incident response workflow for Microsoft 365
Defender

Periodic security operations Example periodic security operations for Microsoft 365
Defender

Investigation for Microsoft Sentinel Incidents in Microsoft Sentinel

Investigation for Microsoft 365 Defender Incidents in Microsoft 365 Defender

Experienced security analyst resources

If you are an experienced security analyst, see these resources to quickly ramp up your SecOps team for
Microsoft security services.

TO P IC RESO URC E

Microsoft Sentinel How to investigate incidents

Microsoft 365 Defender How to investigate incidents

Security operations establishment or modernization Azure Cloud Adoption Framework articles for SecOps and
SecOps functions
 TO P IC RESO URC E

Microsoft security best practices How to best use your SecOps center

Incident response playbooks Overview at https://aka.ms/IRplaybooks

- Phishing
- Password spray
- App consent grant

SOC Process Framework Microsoft Sentinel

MSTICPy and Jupyter Notebooks Microsoft Sentinel

Blog series about SecOps within Microsoft

See this blog series about how the SecOps team at Microsoft works.
Part 1 – Organization: Mission and Culture
Part 2a – People: Teams, Tiers, and Roles
Part 2b – People: Careers and Readiness
Part 3a – Technology: SOC Tooling
Part 3b – Technology: Day in life of an analyst
Part 3c – A day in the life part 2 - Microsoft Security
Part 3d – Zen and the art of threat hunting

Simuland
Simuland is an open-source initiative to deploy lab environments and end-to-end simulations that:
Reproduce well-known techniques used in real attack scenarios.
Actively test and verify the effectiveness of related Microsoft 365 Defender, Microsoft Defender for Cloud,
and Microsoft Sentinel detections.
Extend threat research using telemetry and forensic artifacts generated after each simulation exercise.
Simuland lab environments provide use cases from a variety of data sources including telemetry from Microsoft
365 Defender security products, Microsoft Defender for Cloud, and other integrated data sources through
Microsoft Sentinel data connectors.
In the safety of a trial or paid sandbox subscription, you can:
Understand the underlying behavior and functionality of adversary tradecraft.
Identify mitigations and attacker paths by documenting preconditions for each attacker action.
Expedite the design and deployment of threat research lab environments.
Stay up to date with the latest techniques and tools used by real threat actors.
Identify, document, and share relevant data sources to model and detect adversary actions.
Validate and tune detection capabilities.
The learnings from Simuland lab environment scenarios can then be implemented in your production
environment and security processes.
After reading an overview of Simuland, see the Simuland GitHub repository.
Incident response resources
Planning for your SOC
Process for incident response process recommendations and best practices
Playbooks for detailed guidance on responding to common attack methods
Microsoft 365 Defender incident response
Microsoft Sentinel incident response

Key Microsoft security resources

RESO URC E DESC RIP T IO N

2021 Microsoft Digital Defense Report A report that encompasses learnings from security experts,
practitioners, and defenders at Microsoft to empower people
everywhere to defend against cyberthreats.

Microsoft Cybersecurity Reference Architectures A set of visual architecture diagrams that show Microsoft’s
cybersecurity capabilities and their integration with Microsoft
cloud platforms such as Microsoft 365 and Microsoft Azure
and third-party cloud platforms and apps.

Minutes matter infographic download An overview of how Microsoft's SecOps team does incident
response to mitigate ongoing attacks.

Azure Cloud Adoption Framework security operations Strategic guidance for leaders establishing or modernizing a
security operation function.

Microsoft security best practices for security operations How to best use your SecOps center to move faster than the
attackers targeting your organization.

Microsoft cloud security for IT architects model Security across Microsoft cloud services and platforms for
identity and device access, threat protection, and
information protection.

Microsoft security documentation Additional security guidance from Microsoft.

 Incident response planning
12/13/2021 • 7 minutes to read • Edit Online

Use this table as a checklist to prepare your Security Operations Center (SOC) to respond to cybersecurity
incidents.

DO N E A C T IVIT Y DESC RIP T IO N B EN EF IT

Table top exercises Conduct periodic table top Firmly establishes and
exercises of foreseeable illustrates cybersecurity as a
business-impacting cyber business issue. Develops
incidents that force your muscle memory and
organization's management surfaces difficult decisions
to contemplate difficult risk- and decisions rights issues
based decisions. across the organization.

Determine pre-attack As a compliment to table Defines the initial response

decisions and decision- top exercises, determine parameters and contacts to
makers risk-based decisions, criteria involve that streamline the
for making decisions, and response to an incident.
who must make and
execute those decisions. For
example:

Who/when/if to seek
assistance from law
enforcement?

Who/when/if to enlist
incident responders?

Who/when/if to pay
ransom?

Who/when/if to notify
external auditors?

Who/when/if to notify
privacy regulatory
authorities?

Who/when/if to notify
securities regulators?

Who/when/if to notify
board of directors or audit
committee?

Who has authority to

shut down mission-critical
workloads?
DO N E A C T IVIT Y DESC RIP T IO N B EN EF IT

Maintaining privilege Generally, advice can be Maintaining privilege can be

privileged, but facts are a messy process when
discoverable. Train key considering the multitude
incident leaders in of communications
communicating advice, facts channels, including e-mail,
and opinions under collaboration platforms,
privilege so that privilege is chats, documents, artifacts.
preserved and risk is For example, you can use
reduced. Microsoft Teams Rooms. A
consistent approach across
incident personnel and
supporting external
organizations can help
reduce any potential legal
exposure.

Insider trading Contemplate notifications Boards and external

considerations to management that should auditors tend to appreciate
be taken to reduce that you have mitigations
securities violations risk. that will reduce the risk of
questionable securities
trades during periods of
turbulence.
DO N E A C T IVIT Y DESC RIP T IO N B EN EF IT

Incident roles and Establish basic roles and Technical Incident

responsibilities playbook responsibilities that allow Leader – Always in the
various processes to incident, synthesizing inputs
maintain focus and forward and findings and planning
progress. next actions.

When your response team Communications Liaison

is remote, it can require – Removes the burden of
additional considerations communicating to
for timezones and proper management from the
handoff to investigators. Technical Incident Leader so
they can remain involved in
You might have to the incident without loss of
communicate across other focus.
teams that might be
involved, such as vendor This should include
teams. managing executive
messaging and interactions
and other third parties such
as regulators.

Incident Recorder –
Removes the burden of
recording findings,
decisions, and actions from
an incident responder and
produces an accurate
accounting of the incident
from beginning to end.

For ward Planner –

Working with mission-
critical business process
owners, formulates business
continuity activities and
preparations that
contemplate information
system impairment that
lasts for 24, 48, 72, 96
hours, or more.

Public Relations – In the

event of an incident that is
likely to garner public
attention, and in
conjunction with Forward
Planner, contemplates and
drafts public
communication approaches
that address likely
outcomes.
DO N E A C T IVIT Y DESC RIP T IO N B EN EF IT

Privacy incident response In order to satisfy Evaluating security incidents

playbook increasingly-strict privacy for their potential to impact
regulations, develop a privacy is difficult due to the
jointly-owned playbook fact that most security
between the SecOps and incidents arise in a highly-
the privacy office that technical SOC that must
allows rapid evaluation of quickly get surfaced to a
potential privacy issues that privacy office where
have a reasonable regulatory risk is
probability of arising out of determined, often with a
security incidents. 72-hour notification
expectation.

Penetration testing Conduct point-in-time In light of recent human-

simulated attacks against operated ransomware
business-critical systems, incidents, penetration
critical infrastructure, and testing should be
backups to identify conducted against an
weaknesses in security increased scope of
posture. This is generally infrastructure, particularly
conducted by a team of the ability to attack and
external experts focused on control backups of mission-
bypassing preventative critical systems and data.
controls and surfacing key
vulnerabilities.

Red Team / Blue Team / Conduct continuous or Red, Blue, and Purple team
Purple Team / Green Team periodic simulated attacks attack simulations, when
against business-critical done well, serve a multitude
systems, critical of purposes:
infrastructure, backups to Allows engineers
identify weaknesses in from across the IT
security posture. This is organization to
generally conducted by simulate attacks on
internal attack teams (Red their own
teams) who are focused on infrastructure
testing the effectiveness of disciplines.
detective controls and Surfaces gaps in
teams (Blue teams). visibility and
detection.
For example, you can use Raises the security
Attack simulation training engineering skills
for Microsoft 365 Defender across the board.
for Office 365 and Attack Serves as a more
tutorials & simulations for continuous and
Microsoft 365 Defender for expansive process.
Endpoint.

The Green Team

implements changes in IT or
security configuration.
DO N E A C T IVIT Y DESC RIP T IO N B EN EF IT

Business continuity For mission-critical business Highlights the fact

planning processes, design and test that there is no
continuity processes that continuity
allow the minimum viable workaround for the
business to function during impairment or
times of information absence of IT
systems impairment. systems.
Can emphasize the
For example, use an Azure need and funding
backup and restore plan to for sophisticated
protect your critical digital resilience over
business systems during an simpler backup and
attack to ensure a rapid recovery.
recovery of your business
operations.

Disaster recovery For information systems Organizations that conduct

that support mission-critical bare metal builds often find
business processes, you activities that are impossible
should design and test to replicate or do not fit
hot/cold and hot/warm into the service level
backup and recovery objectives.
scenarios, including staging
times. Mission-critical systems
running on un-supported
hardware many times
cannot be restored to
modern hardware.

Restore of backups is often

not tested and experiences
issues. Backups may be
further offline such that
staging times have not
been factored into recovery
objectives.

Out-of-band Prepare for how you would Although this is a difficult

communications communicate in the event exercise, determine how off-
of e-mail and collaboration line and immutable copies
service impairment, ransom of resources that store
of documentation phone numbers, topologies,
repositories, and build documents, and IT
unavailability of personnel restoration procedures may
phone numbers. be stored on off-line devices
and locations and
distributed at scale.
DO N E A C T IVIT Y DESC RIP T IO N B EN EF IT

Hardening, hygiene, and In line with Center for In response to recent

lifecycle management Internet Security (CIS) Top human-operated
20 security controls, harden ransomware incidents,
your infrastructure and Microsoft has issued specific
perform thorough hygiene guidance for hardening and
activities. protecting every stage of
the cyber attack kill chain
whether with Microsoft
capabilities or those of
other providers. Of
particular note are:
The creation and
maintenance of
immutable backup
copies in the event
of ransomed
systems. You might
also consider how to
keep immutable log
files that complicates
the adversary’s
ability to cover their
tracks.
Risks related to
unsupported
hardware for
disaster recovery.

Incident response planning At the outset of the There is a tendency to

incident, decide on: throw all available resources
Important at an incident in the
organizational beginning and hope for a
parameters. quick resolution. Once you
Assignment of recognize or anticipate that
people to roles and an incident will go for an
responsibilities. extended period of time,
The sense-of- take on a different posture
urgency (such as that with your staff and
24x7 and business suppliers that allows them
hours). to settle in for a longer
Staff for haul.
sustainability for the
duration.
 DO N E A C T IVIT Y DESC RIP T IO N B EN EF IT

Incident responders Establish clear expectations Incident responders come

with one another. A popular with different techniques
format of reporting and approaches, including
ongoing activities includes: dead box analysis, big data
What have we done analysis, and the ability to
(and what were the produce incremental results.
results)? Starting with clear
What are we doing expectations will facilitate
(and what results clear communications.
will be produced and
when)?
What do we plan to
do next (and when is
it realistic to expect
results)?

Incident response resources

Overview for Microsoft security products and resources for new-to-role and experienced analysts
Process for incident response process recommendations and best practices
Playbooks for detailed guidance on responding to common attack methods
Microsoft 365 Defender incident response
Microsoft Sentinel incident response

Key Microsoft security resources

RESO URC E DESC RIP T IO N

2021 Microsoft Digital Defense Report A report that encompasses learnings from security experts,
practitioners, and defenders at Microsoft to empower people
everywhere to defend against cyberthreats.

Microsoft Cybersecurity Reference Architectures A set of visual architecture diagrams that show Microsoft’s
cybersecurity capabilities and their integration with Microsoft
cloud platforms such as Microsoft 365 and Microsoft Azure
and third-party cloud platforms and apps.

Minutes matter infographic download An overview of how Microsoft's SecOps team does incident
response to mitigate ongoing attacks.

Azure Cloud Adoption Framework security operations Strategic guidance for leaders establishing or modernizing a
security operation function.

Microsoft security best practices for security operations How to best use your SecOps center to move faster than the
attackers targeting your organization.

Microsoft cloud security for IT architects model Security across Microsoft cloud services and platforms for
identity and device access, threat protection, and
information protection.

Microsoft security documentation Additional security guidance from Microsoft.

 Incident response process
12/13/2021 • 13 minutes to read • Edit Online

The first step is to have an incident response plan in place that encompasses both internal and external
processes for responding to cybersecurity incidents. The plan should include how to:
Address attacks that vary with the business risk and impact of the incident, which can vary from an isolated
web site that is no longer available to the compromise of administrator-level credentials.
Define the purpose of the response, such as a return to service or to handle legal or public relations aspects
of the attack.
Prioritize the work that needs to get done in terms of how many people should be working on the incident
and their tasks.
See the incident response planning article for a checklist of activities to consider including in your incident
response plan. Once your incident response plan is in place, test it regularly for the most serious types of
cyberattacks to ensure that your organization can respond quickly and efficiently.
Although each organization’s incident response process may be different based on organizational structure and
capabilities, consider the set of recommendations and best practices in this article for responding to security
incidents.
During an incident, it is critical to:
Keep calm
Incidents are extremely disruptive and can become emotionally charged. Stay calm and focus on
prioritizing your efforts on the most impactful actions first.
Do no harm
Confirm that your response is designed and executed in a way that avoids loss of data, loss of business-
critical functionality, and loss of evidence. Avoid decisions can damage your ability to create forensic
timelines, identify root cause, and learn critical lessons.
Involve your legal department
Determine whether they plan to involve law enforcement so you can plan your investigation and
recovery procedures appropriately.
Be careful when sharing information about the incident publicly
Confirm that anything you share with your customers and the public is based on the advice of your legal
department.
Get help when needed
Tap into deep expertise and experience when investigating and responding to attacks from sophisticated
attackers.
Like diagnosing and treating a medical disease, cybersecurity investigation and response for a major incident
requires defending a system that is both:
Critically important (can’t be shut down to work on it).
Complex (typically beyond the comprehension of any one person).
During an incident, you must strike these critical balances:
Speed
Balance the need to act quickly to satisfy stakeholders with the risk of rushed decisions.
Sharing information
Inform investigators, stakeholders, and customers based on the advice of your legal department to limit
liability and avoid setting unrealistic expectations.
This article is designed to lower the risk to your organization for a cybersecurity incident by identifying common
errors to avoid and providing guidance on what actions you can rapidly take that both reduce risk and meet
stakeholder needs.

NOTE
For additional guidance on preparing your organization for ransomware and other types of multi-stage attacks, see
Prepare your recovery plan.

Response best practices

Responding to incidents can be done effectively from both technical and operations perspectives with these
recommendations.

NOTE
For additional detailed industry guidance, see the NIST Computer Security Incident Handling Guide.

Technical
For the technical aspects of incident response, here are some goals to consider:
Try to identify the scope of the attack operation.
Most adversaries use multiple persistence mechanisms.
Identify the objective of the attack, if possible.
Persistent attackers will frequently return for their objective (data/systems) in a future attack.
Here are some useful tips:
Don’t upload files to online scanners
Many adversaries monitor instance count on services like VirusTotal for discovery of targeted malware.
Carefully consider modifications
Unless you face an imminent threat of losing business-critical data—such as deletion, encryption, and
exfiltration—balance the risk of not making the modification with the projected business impact. For
example, temporarily shutting down your organization's internet access may be necessary to protect
business-critical assets during an active attack.
If changes are necessary where the risk of not doing an action is higher than the risk of doing it,
document the action in a change log. Changes made during incident response are focused on disrupting
the attacker and may impact the business adversely. You will need to roll these changes back after the
recovery process.
 Don’t investigate forever
You must ruthlessly prioritize your investigation efforts. For example, only perform forensic analysis on
endpoints that attackers have actually used or modified. In a major incident where an attacker has
administrative privileges, it is practically impossible to investigate all potentially compromised resources
(which may include all organization resources).
Share information
Confirm that all investigation teams, including all internal teams and external investigators or insurance
providers, are sharing their data with each other, based on the advice of your legal department.
Access the right expertise
Confirm that you integrate people with deep knowledge of the systems into the investigation—such as
internal staff or external entities like vendors—not just security generalists.
Anticipate reduced response capability
Plan for 50% of your staff operating at 50% of normal capacity due to situational stress.
A key expectation to manage with stakeholders is that you may never be able to identify the initial attack
because the data required for this may have been deleted before the investigation starts, such as an attacker
covering their tracks by log rolling.
Operations
For security operations aspects of incident response, here are some goals to consider:
Staying focused
Confirm you keep the focus on business-critical data, customer impact, and getting ready for
remediation.
Providing coordination and role clarity
Establish distinct roles for operations in support of the crisis team and confirm that technical, legal, and
communications teams are keeping each other informed.
Keeping your business perspective
You should always consider the impact on business operations by both adversary actions and your own
response actions.
Here are some useful tips:
Consider the Incident Command System (ICS) for crisis management
If you don’t have a permanent organization that manages security incidents, we recommend using the
ICS as a temporary organizational structure to manage the crisis.
Keep ongoing daily operations intact
Ensure that normal security operations are not completely sidelined to support incident investigations.
This work still needs to be done.
Avoid wasteful spending
Many major incidents result in the purchase of expensive security tools under pressure that are never
deployed or used. If you can’t deploy and use a tool during the investigation—which can include hiring
and training for additional staff with the skill sets needed to operate the tool—defer acquisition until after
you finish the investigation.
 Access deep expertise
Confirm you have the ability to escalate questions and issues to deep experts on critical platforms. This
may require access to the operating system and application vendor for business-critical systems and
enterprise-wide components such as desktops and servers.
Establish information flows
Set clear guidance and expectations for the flow of information between senior incident response leaders
and organization stakeholders. See incident response planning for more information.

Recovery best practices

Recovering from incidents can be done effectively from both technical and operations perspectives with these
recommendations.
Technical
For the technical aspects of recovering from an incident, here are some goals to consider:
Don’t boil the ocean
Limit your response scope so that recovery operation can be executed within 24 hours or less. Plan a
weekend to account for contingencies and corrective actions.
Avoid distractions
Defer long-term security investments like implementing large and complex new security systems or
replacing anti-malware solutions until after the recovery operation. Anything that does not have direct
and immediate impact on the current recovery operation is a distraction.
Here are some helpful tips:
Never reset all passwords at once
Password resets should focus first on known compromised accounts based on your investigation and are
potentially administrator or service accounts. If warranted, user passwords should be reset only in a
staged and controlled manner.
Consolidate execution of recovery tasks
Unless you face an imminent threat of losing business-critical data, you should plan a consolidated
operation to rapidly remediate all compromised resources (such as hosts and accounts) versus
remediating compromised resources as you find them. Compressing this time window will make it
difficult for attack operators to adapt and maintain persistence.
Use existing tools
Research and use the capabilities of tools you have already deployed before trying to deploy and learn a
new tool during a recovery.
Avoid tipping off your adversary
As practical, you should take steps to limit the information available to adversaries about the recovery
operation. Adversaries typically have access to all production data and email in a major cybersecurity
incident. But in reality, most attackers don’t have time to monitor all your communications.
Microsoft’s Security Operations Center (SOC) has used a non-production Microsoft 365 tenant for secure
communication and collaboration for members of the incident response team.
Operations
For the operations aspects of recovering from an incident, here are some goals to consider:
Have a clear plan and limited scope
Work closely with your technical teams to build a clear plan with limited scope. While plans may change
based on adversary activity or new information, you should work diligently to limit scope expansion and
taking on additional tasks.
Have clear plan ownership
Recovery operations involve many people doing many different tasks at once, so designate a project lead
for the operation for clear decision-making and definitive information to flow among the crisis team.
Maintain stakeholder communications
Work with communication teams to provide timely updates and active expectation management for
organizational stakeholders.
Here are some helpful tips:
Know your capabilities and limits
Managing major security incidents is very challenging, very complex, and new to many professionals in
the industry. You should consider bringing in expertise from external organizations or professional
services if your teams are overwhelmed or aren’t confident about what to do next.
Capture the lessons learned
Build and continually improve role-specific handbooks for security operations, even if it’s your first
incident without any written procedures.
Executive and board-level communications for incident response can be challenging if not practiced or
anticipated. Make sure you have a communication plan to manage progress reporting and expectations for
recovery.

Incident response process

Consider this general guidance about the incident response process for your security operations and staff.
1. Decide and act
After a threat detection tool such as Microsoft Sentinel or Microsoft 365 Defender detects a likely attack, it
creates an incident. The Mean Time to Acknowledge (MTTA) measurement of SOC responsiveness begins with
the time this attack is noticed by your security staff.
An analyst on shift is either delegated or takes ownership of the incident and performs an initial analysis. The
timestamp for this is the end of the MTTA responsiveness measurement and begins the Mean Time to
Remediate (MTTR) measurement.
As the analyst that owns the incident develops a high enough level of confidence that they understand the story
and scope of the attack, they can quickly shift to planning and executing cleanup actions.
Depending on the nature and scope of the attack, your analysts can clean up attack artifacts as they go (such as
emails, endpoints, and identities) or they may build a list of compromised resources to clean up all at once
(known as a Big Bang)
Clean as you go
For most typical incidents that are detected early in the attack operation, analysts can quickly clean up the
artifacts as they find them. This puts the adversary at a disadvantage and prevents them from moving
forward with the next stage of their attack.
 Prepare for a Big Bang
This approach is appropriate for a scenario where an adversary has already settled in and established
redundant access mechanisms to your environment. This is frequently seen in customer incidents
investigated by Microsoft’s Detection and Response Team (DART). In this approach, analysts should avoid
tipping off the adversary until full discovery of the attacker's presence, because surprise can help with
fully disrupting their operation.
Microsoft has learned that partial remediation often tips off an adversary, which gives them a chance to
react and rapidly make the incident worse. For example, the attacker can spread the attack further, change
their access methods to evade detection, cover their tracks, and inflict data and system damage and
destruction for revenge.
Cleaning up phishing and malicious emails can often be done without tipping off the attacker but
cleaning up host malware and reclaiming control of accounts has a high chance of discovery.
These are not easy decisions to make and there is no substitute for experience in making these judgement calls.
A collaborative work environment and culture in your SOC helps ensure that analysts can tap into each other’s
experience.
The specific response steps are dependent on the nature of the attack, but the most common procedures used
by analysts can include:
Client endpoints (devices)
Isolate the endpoint and contact the user or IT operations/helpdesk to initiate a reinstallation procedure.
Server or applications
Work with IT operations and application owners to arrange rapid remediation of these resources.
User accounts
Reclaim control by disabling the account and resetting password for compromised accounts. These
procedures could evolve as your users transition to passwordless authentication using Windows Hello or
another form of multi-factor authentication (MFA). A separate step is to expire all authentication tokens
for the account with Microsoft Defender for Cloud Apps.
Your analysts can also review the MFA method phone number and device enrollment to ensure it hasn’t
been hijacked by contacting the user and reset this information as needed.
Service Accounts
Because of the high risk of service or business impact, your analysts should work with the service
account owner of record, falling back on IT operations as needed, to arrange rapid remediation of these
resources.
Emails
Delete the attack or phishing email and sometimes clear them to prevent users from recovering deleted
emails. Always save a copy of original email for later search for post-attack analysis, such as headers,
content, and scripts or attachments.
Other
You can execute custom actions based on the nature of the attack such as revoking application tokens and
reconfiguring servers and services.
2. Post-incident cleanup
Because you don’t benefit from learned lessons until you change future actions, always integrate any useful
information learned from the investigation back into your security operations.
Determine the connections between past and future incidents by the same threat actors or methods and capture
these learnings to avoid repeating manual work and analysis delays in the future.
These learnings can take a number of forms, but common practices include analysis of:
Indicators of Compromise (IoCs).
Record any applicable IoCs such as file hashes, malicious IP addresses, and email attributes into your SOC
threat intelligence systems.
Unknown or unpatched vulnerabilities.
Your analysts can initiate processes to ensure that missing security patches get applied, misconfigurations
are corrected, and vendors (including Microsoft) are informed of “zero day” vulnerabilities so that they
can create and distribute security patches.
Internal actions such as enabling logging on assets covering your cloud-based and on-premises
resources.
Review your existing security baselines and consider adding or changing security controls. For example,
see the Azure Active Directory security operations guide for information on enabling the appropriate
level of auditing in the directory before the next incident happens.
Review your response processes to identify and resolve any gaps found during the incident.

Incident response resources

Overview for Microsoft security products and resources for new-to-role and experienced analysts
Planning for your SOC
Playbooks for detailed guidance on responding to common attack methods
Microsoft 365 Defender incident response
Microsoft Sentinel incident response

Key Microsoft security resources

RESO URC E DESC RIP T IO N

2021 Microsoft Digital Defense Report A report that encompasses learnings from security experts,
practitioners, and defenders at Microsoft to empower people
everywhere to defend against cyberthreats.

Microsoft Cybersecurity Reference Architectures A set of visual architecture diagrams that show Microsoft’s
cybersecurity capabilities and their integration with Microsoft
cloud platforms such as Microsoft 365 and Microsoft Azure
and third-party cloud platforms and apps.

Minutes matter infographic download An overview of how Microsoft's SecOps team does incident
response to mitigate ongoing attacks.

Azure Cloud Adoption Framework security operations Strategic guidance for leaders establishing or modernizing a
security operation function.

Microsoft security best practices for security operations How to best use your SecOps center to move faster than the
attackers targeting your organization.
RESO URC E DESC RIP T IO N

Microsoft cloud security for IT architects model Security across Microsoft cloud services and platforms for
identity and device access, threat protection, and
information protection.

Microsoft security documentation Additional security guidance from Microsoft.

 Incident response playbooks
12/13/2021 • 2 minutes to read • Edit Online

You need to respond quickly to detected security attacks to contain and remediate its damage. As new
widespread cyberattacks happen, such as Nobellium and the Exchange Server vulnerability, Microsoft will
respond with detailed incident response guidance.
You also need detailed guidance for common attack methods that malicious users employ every day. To address
this need, use the incident response playbooks for these types of attacks:
Phishing
Password spray
App consent grant
Each playbook includes:
Prerequisites: The specific requirements you need to complete before starting the investigation. For
example, logging that should be turned on and roles and permissions that are required.
Workflow: The logical flow that you should follow to perform the investigation.
Checklist: A list of tasks for the steps in the flow chart. This checklist can be helpful in highly regulated
environments to verify what you have done.
Investigation steps: Detailed step-by-step guidance for the specific investigation.
Also see Microsoft’s DART ransomware approach and best practices for information on how the Detection and
Response Team (DART) at Microsoft deals with ransomware attacks.

Incident response resources

Overview for Microsoft security products and resources for new-to-role and experienced analysts
Planning for your Security Operations Center (SOC)
Process for incident response process recommendations and best practices
Microsoft 365 Defender incident response
Microsoft Sentinel incident response
 Phishing investigation
12/13/2021 • 25 minutes to read • Edit Online

This article provides guidance on identifying and investigating phishing attacks within your organization. The
step-by-step instructions will help you take the required remedial action to protect information and minimize
further risks.
This article contains the following sections:
Prerequisites: Covers the specific requirements you need to complete before starting the investigation. For
example, logging that should be turned on, roles and permissions required, among others.
Workflow: Shows the logical flow that you should follow to perform this investigation.
Checklist: Contains a list of tasks for each of the steps in the flow chart. This checklist can be helpful in
highly regulated environments to verify what you have done or simply as a quality gate for yourself.
Investigation steps: Includes a detailed step-by-step guidance for this specific investigation.

Prerequisites
Here are general settings and configurations you should complete before proceeding with the phishing
investigation.
Account details
Before proceeding with the investigation, it is recommended that you have the user name, user principal name
(UPN) or the email address of the account that you suspect is compromised.
Microsoft 365 base requirements
Verify auditing settings
Verify mailbox auditing on by default is turned on. To make sure that mailbox auditing is turned on for your
organization, run the following command in Microsoft Exchange Online PowerShell:

Get-OrganizationConfig | Format-List AuditDisabled

The value False indicates that mailbox auditing on by default is enabled for the organization. This on by default
organizational value overrides the mailbox auditing setting on specific mailboxes. For example, if mailbox
auditing is disabled for a mailbox (the AuditEnabled property is False on the mailbox), the default mailbox
actions will still be audited for the mailbox, because mailbox auditing on by default is enabled for the
organization.

NOTE
If the tenant was created BEFORE 2019, then you should enable the mailbox auditing and ALL auditing settings. See how
to enable mailbox auditing.

Verify mailbox auditing settings in the tenant

To verify all mailboxes in a given tenant, run the following command in the Exchange Online PowerShell:

Get-Mailbox -ResultSize Unlimited -Filter {RecipientTypeDetails -eq "UserMailbox"} | Set-Mailbox -

AuditEnabled $true
When a mailbox auditing is enabled, the default mailbox logging actions are applied:
AuditLogAgeLimit: 90 days
AuditAdmin:(Actions): Update, Move, MoveToDeletedItems, SoftDelete, HardDelete, FolderBind, SendAs,
SendonBehalf, Create
AuditDelegate:(Actions): Update, SoftDelete, HardDelete, SendAs, Create
AuditOwner: (Actions) {}
To enable the setting for specific users, run the following command. In this example, the user is
[email protected].

Get-Mailbox -Identity "johndoe" | Set-Mailbox -AuditEnabled $true

Message tracing
Message tracing logs are invaluable components to trace message of interest in order to understand the original
source of the message as well as the intended recipients. You can use the MessageTrace functionality through
the Microsoft Exchange Online portal or the Get-MessageTrace PowerShell cmdlet.
Several components of the MessageTrace functionality are self-explanatory but Message-ID is a unique identifier
for an email message and requires thorough understanding. To obtain the Message-ID for an email of interest
we need to examine the raw email headers.
Microsoft 365 security and compliance center
To check whether a user viewed a specific document or purged an item in their mailbox, you can use the Office
365 Security & Compliance Center and check the permissions and roles of users and administrators.
You can also search the unified audit log and view all the activities of the user and administrator in your Office
365 organization.
Are the sign-in logs and/or audit logs exported to an external system?
Since most of the Azure Active Directory (Azure AD) sign-in and audit data will get overwritten after 30 or 90
days, Microsoft recommends that you leverage Sentinel, Azure Monitor or an external SIEM.

Roles and permissions required

Roles in Azure AD
We recommend the following roles are enabled for the account you will use to perform the investigation:
Global Reader
Security Reader
As a last resort, you can always fall back to the role of a Global Administrator / Company Administrator
Roles in Microsoft 365 security and compliance center
Generally speaking, the Global Reader or the Security Reader role should give you sufficient permissions to
search the relevant logs.

NOTE
If a user has the View-Only Audit Logs or Audit Logs role on the Permissions page in the Security & Compliance
Center, they won't be able to search the Office 365 audit log. In this scenario, you must assign the permissions in
Exchange Online because an Exchange Online cmdlet is used to search the log.

If you have implemented the role-based access control (RBAC) in Exchange or if you are unsure which role you
need in Exchange, you can use PowerShell to get the roles required for an individual Exchange PowerShell
cmdlet:

$Perms | foreach {Get-ManagementRoleAssignment -Role $_.Name -Delegating $false | Format-Table -Auto

Role,RoleAssigneeType,RoleAssigneeName}

For more information, see permissions required to run any Exchange cmdlet.
Microsoft Defender for Endpoint
If you have Microsoft Defender for Endpoint (MDE) enabled and rolled out already, you should leverage it for
this flow. See Tackling phishing with signal-sharing and machine learning.

System requirements
Hardware requirements
The system should be able to run PowerShell.
Software requirements
The following PowerShell modules are required for the investigation of the cloud environment:
Azure AD
Azure AD preview in some cases
MS Online for Office 365
Exchange connecting to Exchange for utilizing the unified audit log searches (inbox rules, message traces,
forwarding rules, mailbox delegations, among others)
Azure AD Incident Response
When you use Azure AD commands that are not part of the built-in modules in Azure, you need the MSOnline
module - which is the same module that is used for Office 365. To work with Azure AD (which contains a set of
functions) from PowerShell, install the Azure AD module.

Installing various PowerShell modules

Install the Azure AD PowerShell module
To install the Azure AD PowerShell module, follow these steps:
1. Run the Windows PowerShell app with elevated privileges (run as administrator).
2. To allow PowerShell to run signed scripts, run the following command:

Set-ExecutionPolicy RemoteSigned

3. To install the Azure AD module, run the following command:

Install-Module -Name AzureAD -Verbose

NOTE
If you are prompted to install modules from an untrusted repository, type Y and press Enter .

Install the MSOnline PowerShell module

To install the MSOnline PowerShell module, follow these steps:
1. Run the Windows PowerShell app with elevated privileges (run as administrator).
2. To allow PowerShell to run signed scripts, run the following command:

Set-ExecutionPolicy RemoteSigned

3. To install the MSOnline module, run the following command:

Install-Module -Name MSOnline -Verbose

NOTE
If you are prompted to install modules from an untrusted repository, type Y and press Enter .

Install the Exchange PowerShell module

Please follow the steps on how to get the Exchange PowerShell installed with multi-factor authentication (MFA).
You must have access to a tenant, so you can download the Exchange Online PowerShell module from the
Hybrid tab in the Exchange admin center (EAC).
Once you have configured the required settings, you can proceed with the investigation.
Install the Azure AD Incident Response module
The new AzureADIncidentResponse PowerShell module provides rich filtering capabilities for Azure AD
incidents. Use these steps to install it.
1. Run the Windows PowerShell app with elevated privileges (run as administrator).
2. Use this command.

Install-Module -Name AzureADIncidentResponse -RequiredVersion 4.0

NOTE
If you are prompted to install modules from an untrusted repository, type Y and press Enter .

Workflow
You can also:
Download the phishing and other incident response playbook workflows as a PDF.
Download the phishing and other incident response playbook workflows as a Visio file.

Checklist
This checklist will help you evaluate your investigation process and verify whether you have completed all the
steps during investigation:
Review initial phishing email
Get the list of users who got this email
Get the latest dates when the user had access to the mailbox
Is delegated access configured on the mailbox?
Is there a forwarding rule configured for the mailbox?
Review your Mail Transport Rules
Find the email(s)
Did the user read or open the email?
Who else got the same email?
Did the email contain an attachment?
Was there payload in the attachment?
Check email header for true source of the sender
Verify IP addresses to attackers/campaigns
Did the user click the link in the email?
On what endpoint was the email opened?
Was the attachment payload executed?
Was the destination IP or URL touched or opened?
Was malicious code executed?
What sign-ins happened with the account for the federated scenario?
What sign-ins happened with the account for the managed scenario?
Investigate the source IP address
Investigate the device ID found
Investigate each App ID
You can also download the phishing and other incident playbook checklists as an Excel file.
Investigation steps
For this investigation, it is assumed that you either have a sample phishing email, or parts of it like the sender’s
address, subject of the email, or parts of the message to start the investigation. Please also make sure that you
have completed / enabled all settings as recommended in the Prerequisites section.
This playbook is created with the intention that not all Microsoft customers and their investigation teams will
have the full Microsoft 365 E5 or Azure AD Premium P2 license suite available or configured in the tenant that is
being investigated. We will however highlight additional automation capabilities when appropriate.
Get the list of users / identities who got the email
As the very first step, you need to get a list of users / identities who received the phishing email. The objective of
this step is to record a list of potential users / identities that you will later use to iterate through for additional
investigation steps. Please refer to the Workflow section for a high-level flow diagram of the steps you need to
follow during this investigation.
We do not give any recommendations in this playbook on how you want to record this list of potential users /
identities. Depending on the size of the investigation, you can leverage an Excel book, a CSV file, or even a
database for larger investigations. There are multiple ways to obtain the list of identities in a given tenant, and
here are some examples.
Create a search filter within the security & compliance center
Navigate to the security & compliance center in Microsoft 365 and create a new search filter, using the
indicators you have been provided. Follow the guidance on how to create a search filter.
For a full list of searchable patterns in the security & compliance center, refer to the article on searchable email
properties.
Sample search patterns
The following example query returns messages that were received by users between April 13, 2016 and April
14, 2016 and that contain the words "action" and "required" in the subject line:

(Received:4/13/2016..4/14/2016) AND (Subject:'Action required')

The following example query returns messages that were sent by chatsuwloginsset12345@outlook[.]com and
that contain the exact phrase "Update your account information" in the subject line.

(From:[email protected]) AND (Subject:"Update your account information")

For more details, see how to search for and delete messages in your organization.
Use the Search-Mailbox cmdlet
You can use the Search-mailbox cmdlet to perform a specific search query against a target mailbox of interest
and copy the results to an unrelated destination mailbox.
The following example query searches Jane Smith mailbox for an email that contains the phrase Invoice in the
subject and copies the results to IRMailbox in a folder named "Investigation."

Search-Mailbox -Identity "Jane Smith" -SearchQuery "Subject:Invoice" -TargetMailbox "IRMailbox" -

TargetFolder "Investigation" LogLevel Full

In this example command, the query searches all tenant mailboxes for an email that contains the phrase
"InvoiceUrgent" in the subject and copies the results to IRMailbox in a folder named "Investigation."
 Get-Mailbox | Search-Mailbox -SearchQuery 'InvoiceUrgent vote' -TargetMailbox "IRMailbox" -TargetFolder
"Investigation" -LogLevel Full

Is delegated access configured on the mailbox?

See how to check whether delegated access is configured on the mailbox.
To create this report, run a small PowerShell script that gets a list of all your users. Then, use the Get-
MailboxPermission cmdlet to create a CSV file of all the mailbox delegates in your tenancy.
Look for unusual names or permission grants. If you see something unusual, contact the mailbox owner to
check whether it is legitimate.
Are there forwarding rule (s) configured for the mailbox?
In this step, you need to check each mailbox that was previously identified for forwarding rules or inbox rules.
For forwarding rules, use the following PowerShell command:

Get-Mailbox | select UserPrincipalName,ForwardingSmtpAddress,DeliverToMailboxAndForward | Export-csv

c:\temp\Forwarding.csv -NoTypeInformation

Here's an example query:

Search-UnifiedAuditLog -startdate 12/16/2019 -EndDate 03/16/2019 -ResultSize 5000 -recordtype exchangeadmin

-Operations New-InboxRule |Export-csv NoTypeInformation -Path c:\temp\Inboxrulesoutput.csv

Additionally, you can also utilize the Inbox and Forwarding Rules report in the Office 365 security & compliance
center.
1. Navigate to Dashboard > Report Viewer - Security & Compliance.
2. Look for unusual target locations, or any kind of external addressing.
3. Also look for forwarding rules with unusual key words in the criteria such as all mail with the word
invoice in the subject. Contact the mailbox owner to check whether it is legitimate.
Review inbox rules
Additionally, check for the removal of Inbox rules. As an example, use the following PowerShell commmand:

Search-UnifiedAuditLog -startDate 12/16/2019 -EndDate 03/16/2020 -Operations Remove-InboxRule |Export-CSV

NoTypeInformation -Path c:\temp\removedInboxRules.csv

Look for inbox rules that were removed, consider the timestamps in proximity to your investigations.
Review mail transport rules
There are two ways to obtain the list of transport rules.
1. In the Exchange admin center, navigate to Mail > Flow > Rules .
2. In the Office 365 Security & Compliance Center, navigate to Dashboard Report Viewer > Security &
Compliance - Exchange Transport Rule report and create a report.
The summary view of the report shows you a list of all the mail transport rules you have configured for your
tenancy. When you select any given rule, you'll see details of the rule in a Summar y pane to the right, which
includes the qualifying criteria and action taken when the rule condition matches.
Look for new rules, or rules that have been modified to redirect the mail to external domains. The number of
rules should be relatively small such that you can maintain a list of known good rules. If you a create a new rule,
then you should make a new entry in the Audit report for that event. You can search the report to determine
who created the rule and from where they created it. If you see something unusual, contact the creator to
determine if it is legitimate.
Get the latest dates when the user had access to the mailbox
In the Office 365 security & compliance center, navigate to unified audit log. Under Activities in the drop-down
list, you can filter by Exchange Mailbox Activities .
The capability to list compromised users is available in the Microsoft 365 security & compliance center.
This report shows activities that could indicate a mailbox is being accessed illicitly. It includes created or received
messages, moved or deleted messages, copied or purged messages, sent messages using send on behalf or
send as, and all mailbox sign ins. The data includes date, IP address, user, activity performed, the item affected,
and any extended details.

NOTE
For this data to be recorded, you must enable the mailbox auditing option.

The volume of data included here could be very substantial, so focus your search on users that would have
high-impact if breached. Look for unusual patterns such as odd times of the day, or unusual IP addresses, and
look for patterns such as high volumes of moves, purges, or deletes.
Did the user read / open the email?
There are two main cases here:
Microsoft Exchange Online
Hybrid Exchange with on-premises Exchange servers.
Microsoft Exchange Online
Use the Search-Mailbox cmdlet to perform a specific search query against a target mailbox of interest and copy
the results to an unrelated destination mailbox.
The following example query searches Janes Smith’s mailbox for an email that contains the phrase Invoice in the
subject and copies the results to IRMailbox in a folder named Investigation.

Search-Mailbox -Identity "Jane Smith" -SearchQuery "Subject:Invoice" -TargetMailbox "IRMailbox" -

TargetFolder "Investigation" LogLevel Full

The following sample query searches all tenant mailboxes for an email that contains the phrase InvoiceUrgent in
the subject and copies the results to IRMailbox in a folder named Investigation.

Get-Mailbox | Search-Mailbox -SearchQuery 'InvoiceUrgent vote' -TargetMailbox "IRMailbox" -TargetFolder

"Investigation" -LogLevel Full

Exchange on-premises
Use the Get-MessageTrackingLog cmdlet to search for message delivery information stored in the message
tracking log. Here's an example:

Get-MessageTrackingLog -Server Mailbox01 -Start "03/13/2018 09:00:00" -End "03/15/2018 17:00:00" -Sender
"[email protected]"

For information about parameter sets, see the Exchange cmdlet syntax.
Who else got the same email?
There are two main cases here: You have Exchange Online or Hybrid Exchange with on-premises Exchange
servers. The workflow is essentially the same as explained in the topic
Get the list of users/identities who got the email.
Exchange Online
Use the Search-Mailbox cmdlet to perform a specific search query against a target mailbox of interest and copy
the results to an unrelated destination mailbox.
This sample query searches all tenant mailboxes for an email that contains the subject InvoiceUrgent in the
subject and copies the results to IRMailbox in a folder named Investigation.

Get-Mailbox | Search-Mailbox -SearchQuery "Subject:InvoiceUrgent" -TargetMailbox "IRMailbox" -TargetFolder

"Investigation" -LogLevel Full

Exchange on-premises
Use the Get-MessageTrackingLog cmdlet to search for message delivery information stored in the message
tracking log. Here's an example:

Get-MessageTrackingLog -Server Mailbox01 -Start "03/13/2018 09:00:00" -End "03/15/2018 17:00:00" -

MessageSubject "InvoiceUrgent"

Did the email contain an attachment?

You have two options for Exchange Online:
1. Use the classic Search-Mailbox cmdlet
2. Use the New-ComplianceSearch cmdlet
Exchange Online
Use the Search-Mailbox cmdlet to perform a specific search query against a target mailbox of interest and copy
the results to an unrelated destination mailbox. Here's an example:

Get-Mailbox -ResultSize unlimited | Search-Mailbox -SearchQuery attachment:trojan* -TargetMailbox

"IRMailbox" -TargetFolder "Investigation" -LogLevel Full

The other option is to use the New-ComplianceSearch cmdlet. Here's an example:

New-ComplianceSearch -Name "Investigation" -ExchangeLocation "Research Department" -ContentMatchQuery

"from:[email protected] AND hasattachment:true"

Exchange on-premises
Use the Search-Mailbox cmdlet to search for message delivery information stored in the message tracking log.
Here's an example:

Search-Mailbox -Identity "Jane Smith"-SearchQuery AttachmentNames:attachment_name -TargetMailbox "IRMailbox"

-TargetFolder "Investigation" -LogLevel Full

NOTE
For Exchange 2013, you need CU12 to have this cmdlet running.

Was there payload in the attachment?

In this step, look for potential malicious content in the attachment, for example, PDF files, obfuscated PowerShell,
or other script codes.
The Malware Detections report shows the number of incoming and outgoing messages that were detected as
containing malware for your organization.
To view this report, in the security & compliance center, go to Reports > Dashboard > Malware Detections.
Similar to the Threat Protection Status report, this report also displays data for the past seven days by
default. However, you can choose filters to change the date range for up to 90 days to view the details. (If you
are using a trial subscription, you might be limited to 30 days of data.) To see the details, select View details
table or export the report.
Check email header for true source of the sender
Many of the components of the message trace functionality are self-explanatory but you need to thoroughly
understand about Message-ID. The Message-ID is a unique identifier for an email message.
To obtain the Message-ID for an email of interest, you need to examine the raw email headers. Examination of
the email headers will vary according to the email client being used. However, typically within Office 365, open
the email message and from the Reading pane, select View Original Message to identify the email client. If in
doubt, a simple search on how to view the message headers in the respective email client should provide
further guidance.
You should start by looking at the email headers. For example, in Outlook 365, open the message, navigate to
File > Info > Proper ties :
Properties screen showing email headers
When viewing an email header, it is recommended to copy and paste the header information into an email
header analyzer provided by MXToolbox or Azure for readability.
Headers Routing Information: The routing information provides the route of an email as its being
transferred between computers.
Sender Policy Framework (SPF): An email validation to help prevent/detect spoofing. In the SPF
record, you can determine which IP addresses and domains can send emails on behalf of the domain.
SPF = Pass: The SPF TXT record determined the sender is permitted to send on behalf of a domain.
SPF = Neutral
SPF = Fail: The policy configuration determines the outcome of the message
Sender IP
 SMTP Mail: Validate if this is a legitimate domain
Common Values: Here is a breakdown of the most commonly used and viewed headers, and their
values. This is valuable information and you can use them in the Search fields in Threat Explorer.
From address
Subject
Message ID
To address
Return-path address
Authentication-Results: You can find what your email client authenticated when the email was sent. It
will provide you with SPF and DKIM authentication.
Originating IP: The original IP can be used to determine if the IP is blocklisted and to obtain the geo
location.
Spam Confidence Level (SCL): This determines the probability of an incoming email is spam.
SCL Rating:
-1: Non-spam coming from a safe sender, safe recipient, or safe listed IP address (trusted partner)
0, 1: Non-spam because the message was scanned and determined to be clean
5, 6: Spam
7, 8, 9: High confidence spam
The SPF record is stored within a DNS database and is bundled with the DNS lookup information. You can
manually check the Sender Policy Framework (SPF) record for a domain by using the nslookup command:
1. Open the command prompt (Star t > Run > cmd ).
2. Type the command as: nslookup -type=txt" a space, and then the domain/host name. For example:

nslookup -type=txt domainname.com

NOTE
-all (reject or fail them - don't deliver the email if anything does not match), this is recommended.

Check if DKIM is enabled on your custom domains in Office 365

You need to publish two CNAME records for every domain they want to add the domain keys identified mail
(DKIM). See how to use DKIM to validate outbound email sent from your custom domain.
Check for domain-based message authentication, reporting, and conformance (DMARC )
You can use this feature to validate outbound emails in Office 365.
Verify IP addresses to attackers/campaigns
To verify or investigate IP addresses that have been identified from the previous investigation steps, you can use
any of these options:
VirusTotal
Microsoft Defender for Endpoint
Public Sources:
Ipinfo.io - Has a free option to obtain geo-location
Censys.io - Has a free option to obtain information about what their passive scans of the internet
know
 AbuseIPDB.com - Has a free option that provides some geolocation
Ask Bing and Google - Search on the IP address
URL reputation
You can use any Windows 10 device and Microsoft Edge browser which leverages the SmartScreen technology.
Here are a few third-party URL reputation examples
Trend Micro Site Safety Check
Google Transparency Report
Talos Intelligency
As you investigate the IP addresses and URLs, look for and correlate IP addresses to indicators of compromise
(IOCs) or other indicators, depending on the output or results and add them to a list of sources from the
adversary.
Did the user click the link in the email?
If the user has clicked the link in the email (on-purpose or not), then this action typically leads to a new process
creation on the device itself. Depending on the device this was performed, you need perform device-specific
investigations. For example, Windows vs Android vs iOS. In this article, we have described a general approach
along with some details for Windows-based devices. If you are using Microsoft Defender for Endpoint (MDE),
then you can also leverage it for iOS and soon Android.
You can investigate these events using Microsoft Defender for Endpoint.
1. VPN/proxy logs
Depending on the vendor of the proxy and VPN solutions, you need to check the relevant logs. Ideally
you are forwarding the events to your SIEM or to Microsoft Sentinel.
2. Using Microsoft Defender for Endpoint
This is the best-case scenario, because you can use our threat intelligence and automated analysis to help
your investigation. For more details, see how to investigate alerts in Microsoft Defender for Endpoint.
The Aler t process tree takes alert triage and investigation to the next level, displaying the aggregated
alerts and surrounding evidences that occurred within the same execution context and time period.
3. Windows-based client devices
Make sure you have enabled the Process Creation Events option. Ideally, you should also enable
command-line Tracing Events.
On Windows clients, which have the above-mentioned Audit Events enabled prior to the investigation,
you can check Audit Event 4688 and determine the time when the email was delivered to the user:
On what endpoint was the email opened?
The tasks here are similar to the previous investigation step:
Did the user click the link in the email?
Was the attached payload executed?
The tasks here are similar to the previous investigation step:
Did the user click the link in the email?
Was the destination IP / URL touched or opened?
The tasks here are similar to the previous investigation step:
Did the user click the link in the email?
Was malicious code executed?
The tasks here are similar to the previous investigation step:
Did the user click the link in the email?
What sign-ins happened with the account?
Check the various sign-ins that happened with the account.

Federated scenario
The audit log settings and events differ based on the operating system (OS) Level and the Active Directory
Federation Services (ADFS) Server version.
See the following sections for different server versions.
Server 2012R2
By default, security events are not audited on Server 2012R2. You need to enable this feature on each ADFS
Server in the Farm. In the ADFS Management console and select Edit Federation Ser vice Proper ties .

You also need to enable the OS Auditing Policy .

Open the command prompt, and run the following command as an administrator.
 auditpol.exe /set /subcategory:”Application Generated” /failure:enable /success:enable

For more details, see how to configure ADFS servers for troubleshooting.
You may want to also download the ADFS PowerShell modules from:
GitHub
Microsoft scriptcenter
Server 2016 and newer
By default, ADFS in Windows Server 2016 has basic auditing enabled. With basic auditing, administrators can
see five or less events for a single request. But you can raise or lower the auditing level by using this command:

Set-AdfsProperties -AuditLevel Verbose

For more details, see auditing enhancements to ADFS in Windows server.

If you have Azure AD Connect Health installed, you should also look into the Risky IP report. The failed sign-in
activity client IP addresses are aggregated through Web Application proxy servers. Each item in the Risky IP
report shows aggregated information about failed AD FS sign-in activities that exceed the designated threshold.

For more details, see Risky IP report.

Server 2012R2
Event ID 342 – "The user name or password are incorrect" in the ADFS admin logs.
For the actual audit events, you need to look at the Security events logs and you should look for events with
Event ID 411 for Classic Audit Failure with the source as ADFS Auditing. Also look for Event ID 412 on successful
authentication.
Event ID 411 - SecurityTokenValidationFailureAudit Token validation failed. See inner exception for more
details.
You may need to correlate the Event with the corresponding Event ID 501.
Server 2016 and newer
For the actual audit events you need to look at the security events logs and you should look for events with look
for Event ID 1202 for successful authentication events and 1203 for failures
Example for Event ID1202:
Event ID 1202 FreshCredentialSuccessAudit The Federation Service validated a new credential. See XML for
details.
Example for Event ID 1203:
Event ID 1203 FreshCredentialFailureAudit The Federation Service failed to validate a new credential. See XML
for failure details.
To get the full list of ADFS Event ID per OS Level, refer to GetADFSEventList.
Managed scenario
Check the Azure AD sign-in logs for the user(s) you are investigating.
Navigate to the Azure AD portal > Sign-in screen
Check the sign-in activities
Check the PowerShell function on GitHub
In the Azure AD portal, navigate to the Sign-ins screen and add/modify the display filter for the timeframe you
found in the previous investigation steps as well as add the user name as a filter, as shown in this image.

You can also search using Graph API. For example, filter on User proper ties and get lastSignInDate along
with it. Search for a specific user to get the last signed in date for this user. For example,
https://graph.microsoft.com/beta/users?
$filter=startswith(displayName,'Dhanyah')&$select=displayName,signInActivity

Or you can use the PowerShell command Get-AzureADUserLastSignInActivity to get the last interactive sign-in
activity for the user, targeted by their object ID. This example writes the output to a date and time stamped CSV
file in the execution directory.

Get-AzureADUserLastSignInActivity -TenantId 536279f6-1234-2567-be2d-61e352b51eef -UserObjectId 69447235-

0974-4af6-bfa3-d0e922a92048 -CsvOutput

Or you can use this command from the AzureADIncidentResponse PowerShell module:

Get-AzureADIRSignInDetail -UserId [email protected] -TenantId 536279f6-1234-2567-be2d-61e352b51eef -

RangeFromDaysAgo 29 -RangeToDaysAgo 3

Investigate source IP address

Based on the source IP addresses that you found in the Azure AD sign-in logs or the ADFS/Federation Server log
files, investigate further to know from where the traffic originated.
Managed user
For a managed scenario, you should start looking at the sign-in logs and filter based on the source IP address:

Or you can use this command from the AzureADIncidentResponse PowerShell module:

Get-AzureADIRSignInDetail -IpAddress 1.2.3.4 -TenantId 536279f6-1234-2567-be2d-61e352b51eef -

RangeFromDaysAgo 29 -RangeToDaysAgo 3 -OutGridView

When you look into the results list, navigate to the Device info tab. Depending on the device used, you will get
varying output. Here are a few examples:
Example 1 - Un-managed device (BYOD):

Example 2 - Managed device (Azure AD join or hybrid Azure AD join):

Check for the DeviceID if one is present. You should also look for the OS and the browser or UserAgent string.

Record the CorrelationID, Request ID and timestamp. You should use CorrelationID and timestamp to correlate
your findings to other events.
Federated user/application
Follow the same procedure that is provided for Federated sign-in scenario.
Look for and record the DeviceID, OS Level, CorrelationID, RequestID.
Investigate the identified DeviceID
This step is relevant for only those devices that are known to Azure AD. For example, from the previous steps, if
you found one or more potential device IDs, then you can investigate further on this device. Look for and record
the DeviceID and Device Owner.
Investigate each AppID
The starting point here are the sign-in logs and the app configuration of the tenant or the federation servers'
configuration.
Managed scenario
From the previously found sign-in log details, check the Application ID under the Basic info tab:
Note the differences between the Application (and ID) to the Resource (and ID). The application is the client
component involved, whereas the Resource is the service / application in Azure AD.
With this AppID, you can now perform research in the tenant. Here's an example:

Get-AzureADApplication -Filter "AppId eq '30d4cbf1-c561-454e-bf01-528cd5eafd58'"

ObjectId | AppId | DisplayName

3af6dc4e-b0e5-45ec-8272-56f3f3f875ad 30d4cbf1-c561-454e-bf01-528cd5eafd58 Claims X-Ray

With this information, you can search in the Enterprise Applications portal. Navigate to All Applications and
search for the specific AppID.

Additional incident response playbooks

Examine guidance for identifying and investigating these additional types of attacks:
Password spray
App consent
 Password spray investigation
12/13/2021 • 18 minutes to read • Edit Online

This article provides guidance on identifying and investigating password spray attacks within your organization
and take the required remedial action to protect information and minimize further risks.
This article contains the following sections:
Prerequisites: Covers the specific requirements you need to complete before starting the investigation. For
example, logging that should be turned on, roles and permissions required, among others.
Workflow: Shows the logical flow that you should follow to perform this investigation.
Checklist: Contains a list of tasks for each of the steps in the flow chart. This checklist can be helpful in
highly regulated environments to verify what you have done or simply as a quality gate for yourself.
Investigation steps: Includes a detailed step-by-step guidance for this specific investigation.
Recover y: Contains high-level steps on how to recover/mitigate from a password spray attack.
References: Contains additional reading and reference materials.

Prerequisites
Before starting the investigation, make sure you have completed the setup for logs and alerts and additional
system requirements.
Set up ADFS logging
Event logging on ADFS 2016
By default, the Microsoft Active Directory Federation Services (ADFS) in Windows Server 2016 has a basic level
of auditing enabled. With basic auditing, administrators can see five or less events for a single request.
To view the current auditing level, you can use this PowerShell command:

Get-AdfsProperties

This table contains the auditing levels that are available.

A UDIT L EVEL P O W ERSH EL L SY N TA X DESC RIP T IO N

 A UDIT L EVEL P O W ERSH EL L SY N TA X DESC RIP T IO N

None Set-AdfsProperties -AuditLevel - None Auditing is disabled and no events will

be logged

Basic (Default) Set-AdfsProperties -AuditLevel - Basic No more than 5 events will be logged
for a single request

Verbose Set-AdfsProperties -AuditLevel - All events will be logged. This will log a
Verbose significant amount of information per
request.

To raise or lower the auditing level, use this PowerShell command:

Set-AdfsProperties -AuditLevel

Set up ADFS 2012 R2/2016/2019 security logging

1. Click Star t , navigate to Programs > Administrative Tools , and then click Local Security Policy .
2. Navigate to the Security Settings\Local Policies\User Rights Management folder, and then
double-click Generate security audits .
3. On the Local Security Setting tab, verify that the ADFS service account is listed. If it is not present, click
Add User or Group and add it to the list, and then click OK .
4. To enable auditing, open a command prompt with elevated privileges and run the following command:

auditpol.exe /set /subcategory:"Application Generated" /failure:enable /success:enable

5. Close Local Security Policy .

6. Next, open the ADFS Management snap-in, click Star t , navigate to Programs > Administrative Tools ,
and then click ADFS Management .
7. In the Actions pane, click Edit Federation Ser vice Proper ties .
8. In the Federation Ser vice Proper ties dialog box, click the Events tab.
9. Select the Success audits and Failure audits check boxes.
10. Click OK to finish and save the configuration.
Install Azure AD Connect Health for ADFS
The Azure Active Directory (Azure AD) Connect Health for ADFS agent allows you to have greater visibility into
your federation environment. It provides you with several pre-configured dashboards like usage, performance
monitoring as well as risky IP reports.
To install ADFS Connect Health, go through the requirements for using Azure AD Connect Health, and then
install the Azure ADFS Connect Health Agent.
Set up risky IP alerts
Once Azure AD Connect Health for ADFS is configured, you should set the thresholds for alerting based on what
is suitable for their environment.
Set up SIEM tool alerts on Microsoft Sentinel
To set up SIEM tool alerts, go through the tutorial on out of the box alerting.
SIEM integration into Microsoft Defender for Cloud Apps
Connect the Security Information and Event Management (SIEM) tool to Microsoft Defender for Cloud Apps,
which currently supports Micro Focus ArcSight and generic common event format (CEF).
For more information, see Generic SIEM Integration.
SIEM integration with Graph API
You can connect SIEM with the Microsoft Graph Security API by using any of the following options:
Directly using the suppor ted integration options – Refer to the list of supported integration options
like writing code to directly connect your application to derive rich insights. Leverage samples to get started.
Use native integrations and connectors built by Microsoft par tners – Refer to the Microsoft Graph
Security API partner solutions to use these integrations.
Use connectors built by Microsoft – Refer to the list of connectors that you can use to connect with the
API through a variety of solutions for Security Incident and Event Management (SIEM), Security Response
and Orchestration (SOAR), Incident Tracking and Service Management (ITSM), reporting, and so on.
For more information, see security solution integrations using the Microsoft Graph Security API.
Using Splunk
You can also use the Splunk platform to set up alerts.
Watch this video tutorial on how to create Splunk alerts
For more information, see Splunk alerting manual

Workflow
You can also:
 Download the password spray and other incident response playbook workflows as a PDF.
Download the password spray and other incident response playbook workflows as a Visio file.

Checklist
Investigation triggers
Received a trigger from SIEM, firewall logs, or Azure AD
Azure AD Identity Protection Password Spray feature or Risky IP
Large number of failed sign-ins (Event ID 411)
Spike in Azure AD Connect Health for ADFS
Another security incident (for example, phishing)
Unexplained activity, such as a sign-in from unfamiliar location or a user getting unexpected MFA prompts
Investigation
What is being alerted?
Can you confirm this is a password spray?
Determine timeline for attack.
Determine the IP address(es) of the attack.
Filter on successful sign-ins for this time period and IP address, including successful password but failed MFA
Check MFA reporting
Is there anything out of the ordinary on the account, such as new device, new OS, new IP address used? Use
Defender for Cloud Apps or Azure Information Protection to detect suspicious activity.
Inform local authorities/third parties for assistance.
If you suspect a compromise, check for data exfiltration.
Check associated account for suspicious behavior and look to correlate to other possible accounts and
services as well as other malicious IP addresses.
Check accounts of anyone working in the same office/delegated access - password hygiene (make sure they
are not using the same password as the compromised account)
Run ADFS help
Mitigations
Check the References section for guidance on how to enable features.
Block IP address of attacker (keep an eye out for changes to another IP address)
Changed user’s password of suspected compromise
Enable ADFS Extranet Lockout
Disabled Legacy authentication
Enabled Azure Identity Protection (sign in and user risk policies)
Enabled MFA (if not already)
Enabled Password Protection
Deploy Azure AD Connect Health for ADFS (if not already)
Recovery
Tag bad IP address in Defender for Cloud Apps, SIEM, ADFS and Azure AD
Check for other forms of mailbox persistence such as forwarding rules or additional delegations added
MFA as primary authentication
Configure SIEM integrations with Cloud
Configure Alerting - Identity Protection, ADFS Health Connect, SIEM and Defender for Cloud Apps
Lessons Learnt (include key stakeholders, third parties, communication teams)
Security posture review/improvements
 Plan to run regular attack simulators
You can also download the password spray and other incident playbook checklists as an Excel file.

Investigation steps
Password spray incident response
Let’s understand a few password spray attack techniques before proceeding with the investigation.
Password compromise: An attacker has successfully guessed the user’s password but has not been able to
access the account due to other controls such as multi-factor authentication (MFA).
Account compromise: An attacker has successfully guessed the user’s password and has successfully gained
access to the account.
Environment discovery
Identify authentication type
As the very first step, you need to check what authentication type is used for a tenant/verified domain that you
are investigating.
To obtain the authentication status for a specific domain name, use the Get-MsolDomain PowerShell command.
Here's an example:

Connect-MsolService
Get-MsolDomain -DomainName "contoso.com"

Is the authentication federated or managed?

If the authentication is federated, then successful sign-ins will be stored in Azure AD. The failed sign-ins will be in
their Identity Provider (IDP). For more information, see ADFS troubleshooting and event logging.
If the authentication type is managed, (Cloud only, password hash sync (PHS) or pass-through authentication
(PTA)), then successful and failed sign-ins will be stored in the Azure AD sign-in logs.

NOTE
The Staged Rollout feature allows the tenant domain name to be federated but specific users to be managed. Determine if
any users are members of this group.

Is Azure AD Connect Health enabled for ADFS?

The RiskyIP report will provide suspect IPs and date/time. Notifications should be enabled.
Also check the federated sign-ins investigation from the Phishing playbook
Is the advanced logging enabled in ADFS?
This is a requirement for ADFS Connect Health but it can be enabled independently
See how to enable ADFS Health Connect)
Also check the Federated sign-ins investigation from the Phishing playbook
Are the logs stored in SIEM?
To check whether you are storing and correlating logs in a Security Information and Event Management (SIEM)
or in any other system, check the following:
Log analytics- pre-built queries
Sentinel- pre-built queries
Splunk – pre-built queries
 Firewall logs
UAL if > 30 days
Understanding Azure AD and MFA reporting
It is important that you understand the logs that you are seeing to be able to determine compromise. Below are
our quick guides to understanding Azure AD Sign-Ins and MFA reporting to help with this. Refer to these
articles:
MFA reporting
Understanding Sign Ins
Incident triggers
An incident trigger is an event or a series of events that causes predefined alert to trigger. An example of this is
the number of bad password attempts has gone above your predefined threshold. Below are further examples
of triggers that can be alerted in password spray attacks and where these alerts are surfaced. Incident triggers
include:
Users
IP
User agent strings
Date/time
Anomalies
Bad password attempts

Graph depicting number of bad password attempts

Unusual spikes in activity are key indicators through Azure AD Health Connect (assuming this is installed). Other
indicators are:
 Alerting through SIEM shows a spike when you collate the logs.
Larger than normal log size for ADFS failed sign-ins (this can be an alert in SIEM tool).
Increased amounts of 342/411 event IDs – username or password is incorrect. Or 516 for extranet lockout.
Hit failed authentication request threshold – Risky IP in Azure AD or SIEM tool alert/both 342 and 411 errors
(To be able to view this information, the advanced logging should be turned on.)
Risky IP in Azure AD Health Connect portal
Risky IP will alert when the customized threshold has been reached for bad passwords in an hour and bad
password count in a day as well as extranet lockouts.

Risky IP report data

The details of failed attempts are available in the tabs IP address and extranet lockouts .

IP address and extranet lockouts in the Risky IP report

Detect password spray in Azure Identity Protection
Azure Identity Protection is an Azure AD Premium P2 feature that has a password-spray detection risk alert and
search feature that you can utilize to get additional information or set up automatic remediation.

Details of a password spray attack

Low and slow attack indicators
Low and slow attack indicators are those where thresholds for account lockout or bad passwords are not being
hit. You can detect these indicators through:
 Failures in GAL order
Failures with repetitive attributes (UA, target AppID, IP block/location)
Timing – automated sprays tend to have a more regular time interval between attempts.
Investigation and mitigation

NOTE
You can perform investigation and mitigation simultaneously during sustained/ongoing attacks.

1. Turn advanced logging on ADFS if it is not already turned on.

2. Determine the date and time of start of the attack.
3. Determine the attacker IP address (might be multiple sources and multiple IP addresses) from the
firewall, ADFS, SIEM or Azure AD.
4. Once password spray confirmed, you might have to inform the local agencies (police, third parties,
among others).
5. Collate and monitor the following Event IDs for ADFS:
ADFS 2012 R2
Audit event 403 – user agent making the request
Audit event 411 – failed authentication requests
Audit event 516 – extranet lockout
Audit Event 342 – failed authentication requests
Audit Event 412 - Successful log in
6. To collect the Audit Event 411 - failed authentication requests, use the following script:
 PARAM ($PastDays = 1, $PastHours)
#************************************************
#ADFSBadCredsSearch.ps1
#Version 1.0
#Date: 6-20-2016
#Author: Tim Springston [MSFT]
#Description: This script will parse the ADFS server's (not proxy) security ADFS
#for events which indicate an incorrectly entered username or password. The script can specify a
#past period to search the log for and it defaults to the past 24 hours. Results >#will be placed
into a CSV for
#review of UPN, IP address of submitter, and timestamp.
#************************************************
cls
if ($PastHours -gt 0)
{$PastPeriod = (Get-Date).AddHours(-($PastHours))}
else
{$PastPeriod = (Get-Date).AddDays(-($PastDays)) }
$Outputfile = $Pwd.path + "\BadCredAttempts.csv"
$CS = get-wmiobject -class win32_computersystem
$Hostname = $CS.Name + '.' + $CS.Domain
$Instances = @{}
$OSVersion = gwmi win32_operatingsystem
[int]$BN = $OSVersion.Buildnumber
if ($BN -lt 9200){$ADFSLogName = "AD FS 2.0/Admin"}
else {$ADFSLogName = "AD FS/Admin"}
$Users = @()
$IPAddresses = @()
$Times = @()
$AllInstances = @()
Write-Host "Searching event log for bad credential events..."
if ($BN -ge 9200) {Get-Winevent -FilterHashTable @{LogName= "Security"; >StartTime=$PastPeriod;
ID=411} -ErrorAction SilentlyContinue | Where-Object{$_.Message -match "The user name or password is
incorrect"} | % {
$Instance = New-Object PSObject
$UPN = $_.Properties[2].Value
$UPN = $UPN.Split("-")[0]
$IPAddress = $_.Properties[4].Value
$Users += $UPN
$IPAddresses += $IPAddress
$Times += $_.TimeCreated
add-member -inputobject $Instance -membertype noteproperty -name >"UserPrincipalName" -value $UPN
add-member -inputobject $Instance -membertype noteproperty -name "IP Address" ->value $IPAddress
add-member -inputobject $Instance -membertype noteproperty -name "Time" -value >
($_.TimeCreated).ToString()
$AllInstances += $Instance
$Instance = $null
}
}
$AllInstances | select * | Export-Csv -Path $Outputfile -append -force ->NoTypeInformation
Write-Host "Data collection finished. The output file can be found at >$outputfile`."
$AllInstances = $null

ADFS 2016/2019
Along with the above event IDs, collate the Audit Event 1203 – Fresh Credential Validation Error.
1. Collate all successful sign-ins for this time on ADFS (if federated). A quick sign-in and logout (at the same
second) can be an indicator of a password being guessed successfully and being tried by the attacker.
2. Collate any Azure AD successful or interrupted events for this time-period for both federated and managed
scenarios.
Monitor and collate Event IDs from Azure AD
See how to find the meaning of error logs.
The following Event IDs from Azure AD are relevant:
 50057 - User account was disabled
50055 - Password expired
50072 – User prompted to provide MFA
50074 - MFA required
50079 - user needs to register security info
53003 - User blocked by Conditional Access
53004 - Cannot configure MFA due to suspicious activity
530032 - Blocked by Conditional Access on Security Policy
Sign-In status Success, Failure, Interrupt
Collate event IDs from Sentinel playbook
You can get all the Event IDs from the Sentinel Playbook that is available on GitHub.
Isolate and confirm attack
Isolate the ADFS and Azure AD successful and interrupted sign-in events. These are your accounts of interest.
Block the IP Address ADFS 2012R2 and above for federated authentication. Here's an example:

Set-AdfsProperties -AddBannedIps "1.2.3.4", "::3", "1.2.3.4/16"

Collect ADFS logs

Collect multiple event IDs within a time frame. Here's an example:

Get-WinEvent -ProviderName 'ADFS' | Where-Object { $_.ID -eq '412' -or $_.ID -eq '411' -or $_.ID -eq '342' -
or $_.ID -eq '516' -and $_.TimeCreated -gt ((Get-Date).AddHours(-"8")) }

Collate ADFS logs in Azure AD

Azure AD Sign-In reports include ADFS sign-in activity when you use Azure AD Connect Health. Filter sign-in
logs by Token Issuer Type "Federated".
Here’s an example PowerShell command to retrieve sign-in logs for a specific IP address:

Get-AzureADIRSignInDetail -TenantId b446a536-cb76-4360-a8bb-6593cf4d9c7f -IpAddress 131.107.128.76

Also, search the Azure portal for time frame, IP address and successful and interrupted sign-in as shown in these
images.
Searching for sign-ins within a specific time frame

Searching for sign-ins on a specific IP address

Searching for sign-ins based on the status

You can then download this data as a .csv file for analysis. For more information, see Sign-in activity reports in
the Azure Active Directory portal.
Prioritize findings
It is important to be able to react to the most critical threat. This can be the attacker has successfully obtained
access to an account and therefore can access/exfiltrate data. The attacker has the password but may not be able
to access the account for example they have the password but are not passing the MFA challenge. Also, the
attacker could not be guessing passwords correctly but continuing to try. During analysis, prioritize these
findings:
Successful sign-ins by known attacker IP address
Interrupted sign-in by known attacker IP address
 Unsuccessful sign-ins by known attacker IP address
Other unknown IP address successful sign-ins
Check legacy authentication
Most attacks use legacy authentication. There are a number of ways to determine the protocol of the attack.
1. In Azure AD, navigate to Sign-Ins and filter on Client App.
2. Select all the legacy authentication protocols that are listed.

List of legacy protocols

3. Or if you have an Azure workspace, you can use the pre-built legacy authentication workbook located in
the Azure Active Directory portal under Monitoring and Workbooks .
 Legacy authentication workbook
Block IP address Azure AD for managed scenario (PHS including staging)
1. Navigate to New named locations .

2. Create a CA policy to target all applications and block for this named location only.
Has the user used this operating system, IP, ISP, device, or browser before?
If the user has not used them before and this activity is unusual, then flag the user and investigate all of their
activities.
Is the IP marked as “risky”?
Ensure you record successful passwords but failed multi-factor authentication (MFA) responses, as this activity
indicates that the attacker is getting the password but not passing MFA.
Set aside any account that appears to be a normal sign-in, for example, passed MFA, location and IP not out of
the ordinary.
MFA reporting
It is important to also check MFA logs as an attacker could have successfully guessed a password but be failing
the MFA prompt. The Azure AD MFA logs shows authentication details for events when a user is prompted for
multi-factor authentication. Check and make sure there are no large suspicious MFA logs in Azure AD. For more
information, see how to use the sign-ins report to review Azure AD Multi-Factor Authentication events.
Additional checks
In Defender for Cloud Apps, investigate activities and file access of the compromised account. For more
information, see:
Investigate compromise with Defender for Cloud Apps
Investigate anomalies with Defender for Cloud Apps
Check whether the user has access to additional resources, such as virtual machines (VMs), domain account
permissions, storage, among others.
If data has been breached, then you should inform additional agencies, such as the police.

Immediate remedial actions

1. Change the password of any account that is suspected to have been breached or if the account password has
been discovered. Additionally, block the user. Make sure you follow the guidelines for revoking emergency
access.
2. Mark any account that has been compromised as “compromised” in Azure Identity Protection.
3. Block the IP address of the attacker. Be cautious while performing this action as attackers can use legitimate
VPNs and this could create more risk as they change IP addresses as well. If you are using Cloud
Authentication, then block the IP address in Defender for Cloud Apps or Azure AD. If federated, you need to
block the IP address at the firewall level in front of the ADFS service.
4. Block legacy authentication if it is being used (this action, however, could impact business).
5. Enable MFA if it is not already done.
6. Enable Identity Protection for the user risk and sign-in risk
7. Check the data that has been compromised (emails, SharePoint, OneDrive, apps). See how to use the activity
filter in Defender for Cloud Apps.
8. Maintain password hygiene. For more information, see Azure AD password protection.
9. You can also refer to ADFS Help.

Recovery
Password protection
Implement password protection on Azure AD and on-premises by enabling the custom-banned password lists.
This configuration will prevent users from setting weak passwords or passwords associated with your
organization:
Enabling password protection
For more information, see how to defend against password spray attacks.
Tagging IP address
Tag the IP addresses in Defender for Cloud Apps to receive alerts related to future use:

Tagging IP addresses
In Defender for Cloud Apps, “tag” IP address for the IP scope and set up an alert for this IP range for future
reference and accelerated response.
Setting alerts for a specific IP address
Configure alerts
Depending on your organization needs, you can configure alerts.
Set up alerting in your SIEM tool and look at improving logging gaps. Integrate ADFS, Azure AD, Office 365 and
Defender for Cloud Apps logging.
Configure the threshold and alerts in ADFS Health Connect and Risky IP portal.
Configure threshold settings

Configure notifications
See how to configure alerts in the Identity Protection portal.
Set up sign-in risk policies with either Conditional Access or Identity Protection
Configure Sign-In risk
Configure User Risk
Configure policy alerts in Defender for Cloud Apps

Recommended defenses
Educate end users, key stakeholders, front line operations, technical teams, cyber security and
communications teams
Review security control and make necessary changes to improve or strengthen security control within your
organization
 Suggest Azure AD configuration assessment
Run regular attack simulator exercises

References
Prerequisites
Sentinel Alerting
SIEM integration into Defender for Cloud Apps
SIEM integration with Graph API
Splunk alerting video
Splunk alerting manual
Installing ADFS Health Connect
Understanding Azure AD sign-in logs
Understanding MFA reporting
Mitigations
Mitigations for password spray
Enable password protection
Block legacy authentication
Block IP address on ADFS
Access controls (including blocking IP addresses) ADFS v3
ADFS Password Protection
Enable ADFS Extranet Lockout
MFA as primary authentication
Enable Identity Protection
Azure AD audit activity reference
Azure AD audit logs schema
Azure AD sign-in logs schema
Azure AD audit log Graph API
Risky IP Alerts
ADFS Help
Recovery
SIEM tool integrations
Create Defender for Cloud Apps alerts
Create Risky IP and ADFS Health Connect Alerts
Identity Protection alerts
Attack simulator

Additional incident response playbooks

Examine guidance for identifying and investigating these additional types of attacks:
Phishing
App consent
 App consent grant investigation
12/13/2021 • 21 minutes to read • Edit Online

This article provides guidance on identifying and investigating app consent attacks, protecting information, and
minimizing further risks.
This article contains the following sections:
Prerequisites: Covers the specific requirements you need to complete before starting the investigation. For
example, logging that should be turned on, roles and permissions required, among others.
Workflow: Shows the logical flow that you should follow to perform this investigation.
Checklist: Contains a list of tasks for each of the steps in the flow chart. This checklist can be helpful in
highly regulated environments to verify what you have done or simply as a quality gate for yourself.
Investigation steps: Includes a detailed step-by-step guidance for this specific investigation.
Recover y: Contains high level steps on how to recover/mitigate from an Illicit Application Consent grant
attack.
References: Contains additional reading and reference materials.

Prerequisites
Here are general settings and configurations you should complete to perform an investigation for Application
Consent Grants. Before starting the investigation, make sure you have read about the types of consent
permissions explained in Consent permission types.
Customer data
To start the investigation process, you need the following data:
Access to the tenant as a Global Admin - A Cloud only account (not part of their on-premises environment)
Detail of indicators of compromise (IoCs)
The date and time when you noticed the incident
Date range
Number of compromised accounts
Name(s) of compromised accounts
Roles of the compromised account
Are the accounts highly privileged (GA Microsoft Exchange, SharePoint)?
Are there any Enterprise Applications that are related to the incident?
Did any users report about any applications that were requesting permissions to data on their behalf?
System requirements
Ensure you complete the following installations and configuration requirements:
1. The AzureAD and MSOnline PowerShell modules are installed.
2. You have global administrator rights on the tenant that the script will be run against.
3. You are assigned local administrator role on the computer that you will use to run the scripts.
Install the AzureAD module
Use this command to install the AzureAD module.
 Install-Module -Name AzureAD -Verbose

NOTE
If you are prompted to install the modules from an untrusted repository, type Y and press Enter .

Install the MSOnline PowerShell module

1. Run the Windows PowerShell app with elevated privileges (run as administrator).
2. Run this command to allow PowerShell to run signed scripts.

Set-ExecutionPolicy RemoteSigned

3. Install the MSOnline module with this command.

Install-Module -Name MSOnline -Verbose

NOTE
If you are prompted to install the modules from an untrusted repository, type Y and press Enter .

Download the AzureADPSPermissions Script from GitHub

1. Download the Get-AzureADPSPermissions.ps1 script from GitHub to a folder from which you will run the
script. The output file "permissions.csv" will also be written to this same folder.
2. Open a PowerShell instance as an administrator and open the folder in which you saved the script.
3. Connect to your directory using the Connect-AzureAD cmdlet. Here's an example.

Connect-AzureAD -tenantid "2b1a14ac-2956-442f-9577-1234567890ab" -AccountId

"[email protected]"

4. Run this PowerShell command.

Get-AzureADPSPermissions.ps1 | Export-csv -Path "Permissions.csv" -NoTypeInformation

Disconnect your AzureAD session with this command.

Disconnect-AzureAD

Consent terminologies
What are application consent grants?
Consent is the process of granting authorization to an application to access protected resources on the users’
behalf. An administrator or user can be asked for consent to allow access to their organization/individual data.
An application is granted access to data based on a particular user or for the entire organization. These consents,
however, can be misused by attackers to gain persistence to the environment and access sensitive data. These
types of attacks are called Illicit Consent Grants, which can happen through a phishing email, a user account
compromise through password spray, or when an attacker registers an application as a legitimate user. In
scenarios where a Global Admin account is compromised, then the registration and consent grant are for
tenant-wide and not just for one user.
Before an application can access your organization's data, a user must grant the application permissions to do
so. Different permissions allow different levels of access. By default, all users are allowed to consent to
applications for permissions that don't require administrator consent. For instance, by default, a user can
consent to allow an app to access their mailbox but can't consent to allow an app unfettered access to read and
write to all files in your organization.

NOTE
By allowing users to grant apps access to data, users can easily acquire useful applications and be productive. However, in
some situations, this configuration can represent a risk if it's not monitored and controlled carefully.

Roles that can grant consent on behalf of the organization

To be able to grant tenant-wide admin consent , you must sign in as one of the following:
Global Administrator
Application Administrator
Cloud Application Administrator
Consent types
Administrator - Indicates the consent was provided by the administrator (on behalf of the organization)
Individual user - Indicates the consent was granted by the user and only has access to that user’s
information
Accepted values
AllPrincipals - Consented by an administrator for the entire tenancy
Principal – Consented by the individual user for data only related to that account
Consent and permissions
The actual user experience of granting consent will differ depending on the policies set on the user's tenant, the
user's scope of authority (or role), and the type of permissions being requested by the client application. This
means that application developers and tenant admins have some control over the consent experience. Admins
have the flexibility of setting and deactivating policies on a tenant or app to control the consent experience in
their tenant. Application developers can dictate what types of permissions are being requested and if they want
to guide users through the user consent flow or the admin consent flow.
User consent flow - When an application developer directs users to the authorization endpoint with the
intent to record consent for only the current user.
Admin consent flow - When an application developer directs users to the admin consent endpoint with
the intent to record consent for the entire tenant. To ensure the admin consent flow works properly,
application developers must list all permissions in the RequiredResourceAccess property in the
application manifest.
Delegated permissions vs. application permissions
Delegated permissions are used by apps that have a signed-in user present and can have consents applied by
the administrator or user.
Application permissions are used by apps that run without a signed-in user present. For example, apps that run
as background services or daemons. Application permissions can be consented only by an administrator.
For more information see:
 Admin consent workflow for admin approval for specific applications
Publisher verification program
Configure how end users consent to applications
Classifying risky permissions
There are thousands (at least) of permissions in the system, and not feasible to list out or parse all of these. The
list below will address commonly misused permissions, and others that would create catastrophic impact if
misused.
At a high level, we have observed the following “root” delegated (App+User) permissions being misused in
consent phishing attacks. Root equates to the top level. For example, Contacts.* means to include all delegated
permutations of Contacts permissions: Contacts.Read, Contacts.ReadWrite, Contacts.Read.Shared, and
Contacts.ReadWrite.Shared.
1. Mail.* (including Mail.Send*, but not Mail.ReadBasic*)
2. Contacts. *
3. MailboxSettings.*
4. People.*
5. Files.*
6. Notes.*
7. Directory.AccessAsUser.All
8. User_Impersonation
The first seven permissions in the list above are for Microsoft Graph and the “legacy” API equivalents, such as
Azure Active Directory (Azure AD) Graph and Outlook REST. The eighth permission is for Azure Resource
Manager (ARM), and could also be dangerous on any API that exposes sensitive data with this blanket
impersonation scope.
As per our observation, attackers have used a combination of the first six permissions in the in 99% of the
consent phishing attacks. Most people don’t think of the delegated version of Mail.Read or Files.Read as a high-
risk permission, however, the attacks we’ve seen are generally widespread attacks targeting end users, rather
than spear phishing against admins who can actually consent to the dangerous permissions. It is recommended
to bubble apps with these “critical” level of impact permissions. Even if the applications do not have malicious
intent, and if a bad actor were to compromise the app identity, then your entire organization could be at risk.
For the highest risk impact permissions, star t here:
Application permission (AppOnly/AppRole) versions of all the above permissions, where applicable
Delegated and AppOnly versions of the following permissions:
Application.ReadWrite.All
Directory.ReadWrite.All
Domain.ReadWrite.All*
EduRoster.ReadWrite.All*
Group.ReadWrite.All
Member.Read.Hidden*
RoleManagement.ReadWrite.Directory
User.ReadWrite.All*
User.ManageCreds.All
All other AppOnly permissions that allow write access
For the lowest risk impact permissions list, star t here:
 User.Read
User.ReadBasic.All
Open_id
Email
Profile
Offline_access (only if paired with other permissions on this “lowest risk” list)
Viewing permissions
1. To view the permissions, navigate to the Registration screen in the enterprise application.

2. Select View API permissions .

3. Select Add a permission and the following screen is displayed.

4. Select Microsoft Graph to view the different types of permissions.

5. Select the type of permissions the registered application is using: Delegated permissions or
Application permissions . In the above image, Application permissions is selected.
6. You can search for one of the high-risk impact permissions such as EduRoster .
7. Select EduRoster and expand the permissions.

8. You can now assign or review these permissions.

For more information, read Graph Permissions.

Workflow
You can also:
Download the app consent grant and other incident response playbook workflows as a PDF.
Download the app consent grant and other incident response playbook workflows as a Visio file.

Checklist
Use this checklist to perform application consent grant validation.
Requirements
 Make sure you have access to the tenant as a Global Admin. This is a cloud-only account and is not part
of your on-premises environment.
Indicators of compromise (IoC)
Check the following indicators of compromise (IoC):
When did you notice the incident?
Date range of the incident (how far left is the goal post?)
Number of compromised accounts
Name(s) of compromised accounts
Roles of the compromised account(s)
Are the compromised accounts highly privileged, a standard user, or a combination
Roles
You must be assigned with these roles:
Global administrator rights on the tenant to execute the script
Local Administrator role on the computer from which will run the script
PowerShell configuration
Configure your PowerShell environment with the following:
Install the Azure AD PowerShell module.
Run the Windows PowerShell app with elevated privileges. (Run as administrator).
Configure PowerShell to run signed scripts.
Download the Get-AzureADPSPermissions.ps1 script.
Investigation triggers
Account compromise
App Consent settings modified on the tenant
Alert/audit event status reason "risky application" detected
Noticed odd looking applications
You can also download the app consent grant and other incident playbook checklists as an Excel file.

Investigation steps
You can use the following two methods to investigate application consent grants:
Azure portal
PowerShell script

NOTE
Using the Azure portal will only allow you to see Admin Consent Grants for the last 90 days and based on this, we
recommend using the PowerShell script method only to reduce the attacker registers investigation steps.

Method 1 – Using the Azure portal

You can use the Azure Active Directory portal to find applications to which any individual user has granted
permissions.
1. Sign in to the Azure por tal as an administrator.
2. Select the Azure Active Director y icon.
3. Select Users .
4. Select the user that you want to review.
5. Select Applications .
6. You can see the list of apps that are assigned to the user and what permissions these applications have.
Method 2 - Using PowerShell
There are several PowerShell tools you can use to investigate illicit consent grants, such as:
HAWK tool
AzureAD incident response module
The Get-AzureADPSPermissions.ps1 script from GitHub
PowerShell is the easiest tool and does not require you to modify anything in the tenancy. We are going to base
our investigation on the public documentation from the Illicit Consent Grant attack.
Run Get-AzureADPSPermissions.ps1 , to export all of the OAuth consent grants and OAuth apps for all users in
your tenancy into a .csv file. See the Prerequisites section to download and run the Get-AzureADPSPermissions
script.
1. Open a PowerShell instance as an administrator and open the folder where you saved the script.
2. Connect to your directory using the following Connect-AzureAD command. Here's an example.

Connect-AzureAD -tenantid "2b1a14ac-2956-442f-9577-1234567890ab" -AccountId

"[email protected]"

3. Run this PowerShell command.

Get-AzureADPSPermissions.ps1 | Export-csv c:\temp\consentgrants\Permissions.csv -NoTypeInformation

4. Once the script completes, it is recommended to disconnect the Azure AD session with this command.

Disconnect-AzureAD

NOTE
The script may take hours to complete, depending on the size and permissions configured as well as your
connection.

5. The script creates a file named Permissions.csv.

6. Open the file, filter or format the data into a table and save as an .xlxs file (for filtering).
The column headers for output are shown in this image.

7. In the ConsentType column (G), search for the value AllPrinciples . The AllPrincipals permission
allows the client application to access everyone's content in the tenancy. Native Microsoft 365
applications need this permission to work correctly. Ever y non-Microsoft application with this
permission should be reviewed carefully .
8. In the Permission column (F) , review the permissions that each delegated application has. Look for
Read and Write permission or *. All permission, and review these carefully because they may not be
 appropriate.

NOTE
Review the specific users that have consents granted. If high profile or high impact users have inappropriate
consents granted, you should investigate further.

9. In the ClientDisplayName column (C) , look for apps that seem suspicious, such as:
Apps with misspelled names

Unusual or bland names

Hacker-sounding names. You must review these names carefully.

Example Output: AllPrincipals and read write all. Applications may not have anything suspicious like bland
names and are using MS graph. However, perform research and determine the purpose of the applications and
the actual permissions the applications have in the tenant, as shown in this example.

Here are some useful tips to review information security policy (ISP) investigations:
1. ReplyURL/RedirectURL
Look for suspicious URLs
2. Is the URL hosted on a suspicious domain?
Is it compromised?
Is the domain recently registered?
Is it a temporary domain?
3. Are there terms of service/service agreement link in the app registration?
4. Are the contents unique and specific to the application/publisher?
5. Is the tenant that registered the application either newly created or compromised (for example, is the app
registered by an at-risk user)?

Details of consent grant attack

Attack techniques
While each attack tends to vary, the core attack techniques are:
An attacker registers an app with an OAuth 2.0 provider, such as Azure AD.
The app is configured in a way that makes it seem legitimate. For example, attackers might use the name
of a popular product available in the same ecosystem.
The attacker gets a link directly from users, which may be done through conventional email-based
phishing, by compromising a non-malicious website, or through other techniques.
 The user selects the link and is shown an authentic consent prompt asking them to grant the malicious
app permissions to data.
If a user selects ‘Accept’, they will grant the app permissions to access sensitive data.
The app gets an authorization code, which it redeems for an access token, and potentially a refresh token.
The access token is used to make API calls on behalf of the user.
If the user accepts, the attacker can gain access to the user’s mails, forwarding rules, files, contacts, notes,
profile, and other sensitive data and resources.

Finding signs of an attack

1. Open the Security & Compliance Center.
2. Navigate to Search and select Audit log search .
3. Search (all activities and all users) and enter the start date and end date if required, and then select
Search .

4. Select Filter results and in the Activity field, enter Consent to application.
5. If you have activity under consent to grant, continue as directed below.
6. Select the result to see the details of the activity. Select More Information to get details of the activity.
7. Check whether IsAdminContent is set to ‘True’.

NOTE
This process can take from 30 minutes up to 24 hours for the corresponding audit log entry to be displayed in
the search results after an event occurs.

The extent of time that an audit record is retained and is searchable in the audit log depends on your
Microsoft 365 subscription, and specifically the type of the license that is assigned to a specific user. If
this value is true, it indicates that someone with Global Administrator access may have
granted broad access to data. If this is unexpected, take immediate steps to confirm an
attack .

How to confirm an attack?

If you have one or more instances of the IOCs listed above, you need to do further investigation to positively
confirm that the attack occurred.
Inventor y apps with access in your organization
You can inventory apps for your users using the Azure Active Directory portal, PowerShell, or have your users
individually enumerate their application access.
Use the Azure Active Directory portal to inventory applications and their permissions. This method is
thorough, but you can only check one user at a time, which can be time-consuming if you have to check the
permissions of several users.
Use PowerShell to inventory applications and their permissions. This method is the fastest and most
thorough, with the least amount of overhead.
Encourage your users to individually check their apps and permissions and report the results back to the
administrators for remediation.
Inventory apps assigned to users
You can use the Azure Active Directory portal to see the list of apps to which any individual user has granted
permissions.
1. Sign in to the Azure Por tal with administrative rights.
2. Select the Azure Active Director y icon.
3. Select Users .
4. Select the user that you want to review.
5. Select Applications . You can see the list of apps that are assigned to the user and the permissions granted
to these apps.

Determine the scope of the attack

After you have finished inventorying application access, review the audit log to determine the full scope of the
breach. Search on the affected users, the time frames that the illicit application had access to your organization,
and the permissions the app had. You can search the audit log in the Microsoft 365 Security and Compliance
Center.
Impor tant: If auditing was not enabled prior to the possible attack, you will not be able to investigate as
auditing data will not be available.

How to prevent attacks and mitigate risks?

Regularly audit applications and granted permissions in your organization to ensure no unwarranted or
suspicious applications have previously been granted access to data.
Review, detect, and remediate illicit consent grants in Office 365 for additional best practices and
safeguards against suspicious applications requesting OAuth consent.
If your organization has the appropriate license:
Use additional OAuth application auditing features in Microsoft Defender for Cloud Apps.
Use Azure Monitor Workbooks to monitor permissions and consent related activity. The Consent Insights
workbook provides a view of apps by number of failed consent requests. This can be helpful to prioritize
applications for administrators to review and decide whether to grant them admin consent.

How to stop and remediate an illicit consent grant attack?

After you have identified an application with illicit permissions, you have several ways to remove that access.
You can revoke the application's permission in the Azure Active Director y por tal.
1. Navigate to the affected user in the Azure Active Director y User tab.
2. Select Applications .

3. Select the illicit application.

4. Select Remove .
You can use PowerShell to revoke the OAuth consent grant by following the steps in Remove-
AzureADOAuth2PermissionGrant .
First, run this command to gather information that you have in Azure AD, for permissions of consent grants.

Get-ADOAuth2PermissionGrantoAuth

Here's an example of the output.

You can use PowerShell to revoke the Ser vice App Role Assignment by following the steps in
Remove-AzureADSer viceAppRoleAssignment .
Here's an example.

Remove-AzureADOAuth2PermissionGrant -ObjectId "GbrSwpsCB0ar6c7N7PRvD1bNACUj4C9IspcKu5YkdoE"

Here's an example of the output.

You can also deactivate sign-in for the affected account altogether, which will in turn deactivate app access to
data in that account. This option isn't ideal for the end user's productivity, but it can be a viable short-term
remediation.

Recommended defenses
Steps to protect your organization
There are various consent attack types, but if you follow these recommended defenses, which will mitigate all
types of attacks, especially consent phishing, where attackers trick users into granting a malicious app access to
sensitive data or other resources. Instead of trying to steal the user’s password, an attacker is seeking
permission for an attacker-controlled app to access valuable data.
To help prevent consent attacks from affecting Azure AD and Office 365, see the following recommendations:
Set policies
This setting will have user implications and may not be applicable for an environment. If you are going to
allow any consents, ensure the administrators approve the requests.
Allow consents for applications from verified publishers only and specific types of permissions classified
as low impact.
 NOTE
The above recommendations are suggested based on the most ideal, secure configurations. However, as security
is a fine balance between functionalities and operations, the most secure configurations might cause additional
overheads to administrators. It is a decision best made after consulting with your administrators.

Configure risk-based step-up consent - Enabled by default if user consent to grants is

enabled
Risk-based step-up consent helps reduce user exposure to malicious apps that make illicit consent
requests. If Microsoft detects a risky end-user consent request, the request will require a "step-up" to
admin consent instead. This capability is enabled by default , but it will only result in a behavior change
when end-user consent is enabled .
When a risky consent request is detected, the consent prompt will display a message indicating that
admin approval is needed. If the admin consent request workflow is enabled, the user can send the
request to the admin for further review directly from the consent prompt. If it's not enabled, the following
message is displayed:
AADSTS90094: <clientAppDisplayName> needs permission to access resources in your organization that
only an admin can grant. Please ask an admin to grant permission to this app before you can use it. In
this case, an audit event will also be logged with a Category of "ApplicationManagement" Activity
Type of "Consent to application" , and Status Reason of "Risky application detected" .

NOTE
Any tasks that require administrator’s approval will have operational overhead. The "Consent and permissions, User
consent settings " is in Preview currently. Once it is ready for general availability (GA), the "Allow user consent from
verified publishers, for selected permissions " feature should reduce administrators’ overhead and it is
recommended for most organizations.

Educate your application developers to follow the trustwor thy app ecosystem.
To help developers build high-quality and secure integrations, we’re also announcing public preview of the
Integration Assistant in Azure AD app registrations.
The Integration Assistant analyzes your app registration and benchmarks it against a set of recommended
security best practices.
 The Integration Assistant highlights best practices that are relevant during each phase of your integration’s
lifecycle—from development all the way to monitoring—and ensures every stage is properly configured.
It’s designed to make your job easier, whether you’re integrating your first app or you’re an expert looking to
improve your skills.
Educate your organization on consent tactics ( phishing tactics, admin and user consents ):
Check for poor spelling and grammar. If an email message or the application’s consent screen has spelling
and grammatical errors, it’s likely to be a suspicious application.
Keep a watchful eye on app names and domain URLs. Attackers like to spoof app names that make it appear
to come from legitimate applications or companies but drive you to consent to a malicious app.
Make sure you recognize the app name and domain URL before consenting to an application.
Promote and allow access to apps you trust
Promote the use of applications that have been publisher verified. Publisher verification helps admins and
end users understand the authenticity of application developers. Over 660 applications by 390 publishers
have been verified thus far.
Configure application consent policies by allowing users to only consent to specific applications you trust,
such as applications developed by your organization or from verified publishers.
Educate your organization on how our permissions and consent framework works.
Understand the data and permissions an application is asking for and understand how permissions and
consent work within our platform.
Ensure administrators know how to manage and evaluate consent requests.
Audit apps and consented permissions in your organization to ensure applications being used are accessing
only the data they need and adhering to the principles of least privilege.

Mitigations
Educate the customer and provide awareness and training on securing application consent grants
Tighten the application consent grants process with organizational policy and technical controls
Set up Create schedule to review Consented applications
You can use PowerShell to revoke the OAuth consent grant by following the steps in Remove-
AzureADOAuth2PermissionGrant.
You can use PowerShell to revoke the Service App Role assignment by following the steps in Remove-
AzureADServiceAppRoleAssignment.
You can also deactivate sign-in for the affected account altogether, which will in turn deactivate application
access to data in that account.
You can turn off integrated applications for your tenancy. This is a drastic step that prevents end users from
granting consent to third-party applications on a tenant-wide basis. However, this option is not
recommended.

References
The source of the content for this article is the following:
Protecting remote workforce application attacks
Fostering a secure and trustworthy app ecosystem
Investigate risky OAuth apps
Managing consent to applications and evaluating consent requests
Disable user sign-ins for an enterprise app in Azure Active Directory
Understand the permissions and consent framework in the Microsoft identity platform.
 Understand the difference between delegated permissions and application permissions.
Configure how end-users consent to applications
Unexpected application in my applications list
Detect and Remediate Illicit Consent Grants
How and Why Azure AD Applications are Added
Application and service principal objects in Azure Active Directory
Azure AD Config Documentor
Managing consent to applications and evaluating consent requests
Get-AzureADServicePrincipal
Build 2020: Fostering a secure and trustworthy app ecosystem for all users
Configure the admin consent workflow
Admins should evaluate all consent requests carefully before approving a request, especially when Microsoft
has detected risk.
Application Registration vs. Enterprise Applications
Permissions
KrebsOnSecurity on AppConsent Phishing

Additional incident response playbooks

Examine guidance for identifying and investigating these additional types of attacks:
Phishing
Password spray
 Microsoft’s DART ransomware approach and best
practices
12/13/2021 • 14 minutes to read • Edit Online

Human-operated ransomware is not a malicious software problem—it’s a human criminal problem. The
solutions used to address commodity problems aren’t enough to prevent a threat that more closely resembles a
nation-state threat actor who:
Disables or uninstalls your antivirus software before encrypting files
Disables security services and logging to avoid detection
Locates and corrupts or deletes backups before sending a ransom demand
These actions are commonly done with legitimate programs that you might already have in your environment
for administrative purposes. In criminal hands, these tools are used maliciously to carry out attacks.
Responding to the increasing threat of ransomware requires a combination of modern enterprise configuration,
up-to-date security products, and the vigilance of trained security staff to detect and respond to the threats
before data is lost.
The Microsoft Detection and Response Team (DART) responds to security compromises to help customers
become cyber-resilient. DART provides onsite reactive incident response and remote proactive investigations.
DART leverages Microsoft’s strategic partnerships with security organizations around the world and internal
Microsoft product groups to provide the most complete and thorough investigation possible.
This article describes how DART handles ransomware attacks for Microsoft customers so that you can consider
applying elements of their approach and best practices for your own security operations playbook.
See these sections for the details:
How DART uses Microsoft security services
The DART approach to conducting ransomware incident investigations
DART recommendations and best practices

NOTE
This article content was derived from the A guide to combatting human-operated ransomware: Part 1 and A guide to
combatting human-operated ransomware: Part 2 Microsoft Security team blog posts.

How DART uses Microsoft security services

DART relies heavily on data for all investigations and uses existing deployments of Microsoft security services
such as Microsoft Defender for Office 365, Microsoft Defender for Endpoint, Microsoft Defender for Identity, and
Microsoft Defender for Cloud Apps.
Defender for Endpoint
Defender for Endpoint is Microsoft’s enterprise endpoint security platform designed to help enterprise network
security analysts prevent, detect, investigate, and respond to advanced threats. Defender for Endpoint can detect
attacks using advanced behavioral analytics and machine learning. Your analysts can use Defender for Endpoint
for attacker behavioral analytics.
Here's an example of an alert in Microsoft Defender for Endpoint for a pass-the-ticket attack.
Your analysts can also perform advanced hunting queries to pivot off indicators of compromise (IOCs) or search
for known behavior if they identify a threat actor group.
Here’s an example of how advanced hunting queries can be used to locate known attacker behavior.

In Defender for Endpoint, you have access to a real-time expert-level monitoring and analysis service by
Microsoft Threat Experts for ongoing suspected actor activity. You can also collaborate with experts on demand
for additional insights into alerts and incidents.
Here’s an example of how Defender for Endpoint shows detailed ransomware activity.
Defender for Identity
You use Defender for Identity to investigate known compromised accounts and to find potentially compromised
accounts in your organization. Defender for Identity sends alerts for known malicious activity that actors often
use such as DCSync attacks, remote code execution attempts, and pass-the-hash attacks. Defender for Identity
enables you to pinpoint suspect activity and accounts to narrow down the investigation.
Here's an example of how Defender for Identity sends alerts for known malicious activity related to ransomware
attacks.

Defender for Cloud Apps

Defender for Cloud Apps (previously known as Microsoft Defender for Cloud Apps) allows your analysts to
detect unusual behavior across cloud apps to identify ransomware, compromised users, or rogue applications.
Defender for Cloud Apps is Microsoft’s cloud access security broker (CASB) solution that allows for monitoring
of cloud services and data access in cloud services by users.
Here's an example of the Defender for Cloud Apps dashboard, which allows analysis to detect unusual behavior
across cloud apps.
Microsoft Secure Score
The set of Microsoft 365 Defender services provides live remediation recommendations to reduce the attack
surface. Microsoft Secure Score is a measurement of an organization’s security posture, with a higher number
indicating that more improvement actions have been taken. See the Secure Score documentation to find out
more about how your organization can leverage this feature to prioritize remediation actions that are based on
their environment.

The DART approach to conducting ransomware incident

investigations
You should make every effort to determine how the adversary gained access to your assets so that
vulnerabilities can be remediated. Otherwise, it is highly likely that the same type of attack will take place again
in the future. In some cases, the threat actor takes steps to cover their tracks and destroy evidence, so it is
possible that the entire chain of events may not be evident.
The following are three key steps in DART ransomware investigations:

ST EP GO A L IN IT IA L Q UEST IO N S

1. Assess the current situation Understand the scope What initially made you aware of a
ransomware attack?

What time/date did you first learn of

the incident?

What logs are available and is there

any indication that the actor is
currently accessing systems?
 ST EP GO A L IN IT IA L Q UEST IO N S

2. Identify the affected line-of-business Get systems back online Does the application require an
(LOB) apps identity?

Are backups of the application,

configuration, and data available?

Are the content and integrity of

backups regularly verified using a
restore exercise?

3. Determine the compromise Remove attacker control from the N/A

recovery (CR) process environment

Step 1. Assess the current situation

An assessment of the current situation is critical to understanding the scope of the incident and for determining
the best people to assist and to plan and scope the investigation and remediation tasks. Asking the following
initial questions is crucial in helping to determine the situation.
What initially made you aware of the ransomware attack?
If the initial threat was identified by IT staff—such as noticing backups being deleted, antivirus alerts, endpoint
detection and response (EDR) alerts, or suspicious system changes—it is often possible to take quick decisive
measures to thwart the attack, typically by disabling all inbound and outbound Internet communication. This
may temporarily affect business operations, but that would typically be much less impactful than an adversary
deploying ransomware.
If the threat was identified by a user call to the IT helpdesk, there may be enough advance warning to take
defensive measures to prevent or minimize the effects of the attack. If the threat was identified by an external
entity (like law enforcement or a financial institution), it is likely that the damage is already done, and you will
see evidence in your environment that the threat actor has already gained administrative control of your
network. This can range from ransomware notes, locked screens, or ransom demands.
What date/time did you first learn of the incident?
Establishing the initial activity date and time is important because it helps narrow the scope of the initial triage
for quick wins by the attacker. Additional questions may include:
What updates were missing on that date? This is important to understand what vulnerabilities may have
been exploited by the adversary.
What accounts were used on that date?
What new accounts have been created since that date?
What logs are available, and is there any indication that the actor is currently accessing systems?
Logs—such as antivirus, EDR, and virtual private network (VPN)—are an indicator of suspected compromise.
Follow-up questions may include:
Are logs being aggregated in a Security Information and Event Management (SIEM) solution—such as
Microsoft Sentinel, Splunk, ArcSight, and others—and current? What is the retention period of this data?
Are there any suspected compromised systems that are experiencing unusual activity?
Are there any suspected compromised accounts that appear to be actively used by the adversary?
Is there any evidence of active command and controls (C2s) in EDR, firewall, VPN, web proxy, and other logs?
As part of assessing the current situation, you might need an Active Directory Domain Services (AD DS) domain
controller that was not compromised, a recent backup of a domain controller, or a recent domain controller
taken offline for maintenance or upgrades. Also determine whether multifactor authentication (MFA) was
required for everyone in the company and if Azure Active Directory (Azure AD) was used.
Step 2. Identify the LOB apps that are unavailable due to the incident
This step is critical in figuring out the quickest way to get systems back online while obtaining the evidence
required.
Does the application require an identity?
How is authentication performed?
How are credentials such as certificates or secrets stored and managed?
Are tested backups of the application, configuration, and data available?
Are the contents and integrity of backups regularly verified using a restore exercise? This is particularly
important after configuration management changes or version upgrades.
Step 3. Determine the compromise recovery process
This step may be necessary if you have determined that the control plane, which is typically AD DS, has been
compromised.
Your investigation should always have a goal of providing output that feeds directly into the CR process. CR is
the process that removes attacker control from an environment and tactically increase security posture within a
set period. CR takes place post-security breach. To learn more about CR, read the Microsoft Compromise
Recovery Security Practice team’s CRSP: The emergency team fighting cyber attacks beside customers blog
article.
Once you have gathered the responses to the questions above, you can build a list of tasks and assign owners. A
key factor in a successful incident response engagement is thorough, detailed documentation of each work item
(such as the owner, status, findings, date, and time), making the compilation of findings at the end of the
engagement a straightforward process.

DART recommendations and best practices

Here are DART’s recommendations and best practices for containment and post-incident activities.
Containment
Containment can only happen once the analysis has determined what needs to be contained. In the case of
ransomware, the adversary’s goal is to obtain credentials that allow administrative control over a highly
available server and then deploy the ransomware. In some cases, the threat actor identifies sensitive data and
exfiltrates it to a location they control.
Tactical recovery will be unique for your organization’s environment, industry, and level of IT expertise and
experience. The steps outlined below are recommended for short-term and tactical containment steps your
organization can take. To learn more about for long-term guidance, see securing privileged access. For a
comprehensive view of ransomware and extortion and how to prepare and protect your organization, see
Human-operated ransomware.
These containment steps can be done concurrently as new threat vectors are discovered:
Step 1: Assess the scope of the situation
Which user accounts were compromised?
Which devices are affected?
Which applications are affected?
Step 2: Preserve existing systems
Disable all privileged user accounts except for a small number of accounts used by your admins to
 assist in resetting the integrity of your AD DS infrastructure. If a user account is believed to be
compromised, disable it immediately.
Isolate compromised systems from the network, but do not shut them off.
Isolate at least one known good domain controller in every domain—two is even better. Either
disconnect them from the network or shut them down entirely. The object here is to stop the
spread of ransomware to critical systems—identity being among the most vulnerable. If all your
domain controllers are virtual, ensure that the virtualization platform’s system and data drives are
backed up to offline external media that is not connected to the network, in case the virtualization
platform itself is compromised.
Isolate critical known good application servers, for example SAP, configuration management
database (CMDB), billing, and accounting systems.
These two steps can be done concurrently as new threat vectors are discovered. Disable those threat vectors and
then try to find a known good system to isolate from the network.
Other tactical containment actions can include:
Reset the krbtgt password, twice in rapid succession. Consider using a scripted, repeatable process. This
script enables you to reset the krbtgt account password and related keys while minimizing the likelihood
of Kerberos authentication issues being caused by the operation. To minimize potential issues, the krbtgt
lifetime can be reduced one or more times prior to the first password reset so that the two resets are
done quickly. Note that all domain controllers that you plan to keep in your environment must be online.
Deploy a Group Policy to the entire domain(s) that prevents privileged login (Domain Admins) to
anything but domain controllers and privileged administrative-only workstations (if any).
Install all missing security updates for operating systems and applications. Every missing update is a
potential threat vector that adversaries can quickly identify and exploit. Microsoft Defender for Endpoint’s
Threat and Vulnerability Management provides an easy way to see exactly what is missing—as well as the
potential impact of the missing updates.
For Windows 10 (or higher) devices, confirm that the current version (or n-1) is running on every
device.
Apply attack surface reduction (ASR) rules to prevent malware infection.
Enable all Windows 10 security features.
Check that every external facing application, including VPN access, is protected by multifactor
authentication, preferably using an authentication application that is running on a secured device.
For devices not using Defender for Endpoint as their primary antivirus software, run a full scan with
Microsoft Safety Scanner on isolated known good systems before reconnecting them to the network.
For any legacy operating systems, upgrade to a supported OS or decommission these devices. If these
options are not available, take every possible measure to isolate these devices, including network/VLAN
isolation, Internet Protocol security (IPsec) rules, and login restrictions, so they are only accessible to the
applications by the users/devices to provide business continuity.
The riskiest configurations consist of running mission critical systems on legacy operating systems as old as
Windows NT 4.0 and applications, all on legacy hardware. Not only are these operating systems and
applications insecure and vulnerable, if that hardware fails, backups typically cannot be restored on modern
hardware. Unless replacement legacy hardware is available, these applications will cease to function. Strongly
consider converting these applications to run on current operating systems and hardware.
Post-incident activities
DART recommends implementing the following security recommendations and best practices after each
incident.
Ensure that best practices are in place for email and collaboration solutions to make it more difficult for
attackers to abuse them while allowing internal users to access external content easily and safely.
Follow Zero Trust security best practices for remote access solutions to internal organizational resources.
Starting with critical impact administrators, follow best practices for account security including using
passwordless authentication or MFA.
Implement a comprehensive strategy to reduce the risk of privileged access compromise.
For cloud and forest/domain administrative access, use Microsoft’s privileged access model (PAM).
For endpoint administrative management, use the local administrative password solution (LAPS).
Implement data protection to block ransomware techniques and to confirm rapid and reliable recovery
from an attack.
Review your critical systems. Check for protection and backups against deliberate attacker erasure or
encryption. It’s important that you periodically test and validate these backups.
Ensure rapid detection and remediation of common attacks on endpoint, email, and identity.
Actively discover and continuously improve the security posture of your environment.
Update organizational processes to manage major ransomware events and streamline outsourcing to
avoid friction.
PAM
Using the PAM (formerly known as the tiered administration model) enhances Azure AD’s security posture. This
involves:
Breaking out administrative accounts in a “planed” environment—one account for each level, usually
four:
Control Plane (formerly Tier 0): Administration of domain controllers and other crucial identity services,
such as Active Directory Federation Services (ADFS) or Azure AD Connect. This also includes server
applications that require administrative permissions to AD DS, such as Exchange Server.
The next two planes were formerly Tier 1:
Management Plane: Asset management, monitoring, and security.
Data/Workload Plane: Applications and application servers.
The next two planes were formerly Tier 2:
User Access: Access rights for users (such as accounts).
App Access: Access rights for applications.
Each one of these planes will have a separate administrative workstation for each plane and will only
have access to systems in that plane. Other accounts from other planes will be denied access to
workstations and servers in the other planes through user rights assignments set to those machines.
The net result of the PAM is that:
A compromised user account will only have access to the plane to which it belongs.
More sensitive user accounts will not be logging into workstations and servers with a lower plane’s
 security level, thereby reducing lateral movement.
LAPS
By default, Microsoft Windows and AD DS have no centralized management of local administrative accounts on
workstations and member servers. This usually results in a common password that is given for all these local
accounts, or at the very least in groups of machines. This enables would-be attackers to compromise one local
administrator account, and then use that account to gain access to other workstations or servers in the
organization.
Microsoft’s LAPS mitigates this by using a Group Policy client-side extension that changes the local
administrative password at regular intervals on workstations and servers according to the policy set. Each of
these passwords are different and stored as an attribute in the AD DS computer object. This attribute can be
retrieved from a simple client application, depending on the permissions assigned to that attribute.
LAPS requires the AD DS schema to be extended to allow for the additional attribute, the LAPS Group Policy
templates to be installed, and a small client-side extension to be installed on every workstation and member
server to provide the client-side functionality.
You can get LAPS from the Microsoft Download Center.

Incident response playbooks

Examine guidance for identifying and investigating these types of attacks:
Phishing
Password spray
App consent grant

Incident response resources

Overview for Microsoft security products and resources for new-to-role and experienced analysts
Planning for your Security Operations Center (SOC)
Process for incident response process recommendations and best practices
Microsoft 365 Defender incident response
Microsoft Sentinel incident response
 Microsoft Security Best Practices module: Identity
and access management
12/13/2021 • 2 minutes to read • Edit Online

Identity and access management is critical to both security assurances as an access control as well as enterprise
enablement of applications and services.
The following videos provide guidance on identity and access management.

Part 1: Introduction - Identity Attacks & Key Capabilities (5:44 long)

Part 2: Consistency (4:20 long)

Part 3: Critical Best Practices (4:24 long)

Part 4: Password (Hash) Synchronization with Cloud (4:08 long)

Part 5: Password Protection from Cloud (3:00 long)

Part 6: General Guidance (2:38 long)

Next steps
See the Identity and access management and Capabilities topics.

See also
PowerPoint slides for the Microsoft Azure Security Compass Workshop
Zero Trust Security Model and Framework
Microsoft security documentation
 Microsoft security best practices for identity and
access management
12/13/2021 • 8 minutes to read • Edit Online

In cloud-based architecture, identity provides the basis of a large percentage of security assurances. While
legacy IT infrastructure often heavily relied on firewalls and network security solutions at the internet egress
points for protection against outside threats, these controls are less effective in cloud architectures with shared
services being accessed across cloud provider networks or the internet.
It is challenging or impossible to write concise firewall rules when you don’t control the networks where these
services are hosted, different cloud resources spin up and down dynamically, cloud customers may share
common infrastructure, and employees and users expect to be able to access data and services from anywhere.
To enable all these capabilities, you must manage access based on identity authentication and authorization
controls in the cloud services to protect data and resources and to decide which requests should be permitted.
Additionally, using a cloud-based identity solution like Azure Active Directory (Azure AD) offers additional
security features that legacy identity services cannot because they can apply threat intelligence from their
visibility into a large volume of access requests and threats across many customers.

A single enterprise directory

Best practice: Establish a single enterprise directory for managing identities of full-time employees and
enterprise resources.
A single authoritative source for identities increases clarity and consistency for all roles in IT and Security. This
reduces security risk from human errors and automation failures resulting from complexity. By having a single
authoritative source, teams that need to make changes to the directory can do so in one place and have
confidence that their change will take effect everywhere.
For Azure, designate a single Azure AD tenant as the authoritative source for your organization's accounts.
For more information, see What is hybrid identity?.

Synchronized identity systems

Best practice: Synchronize your cloud identity with your existing identity systems.
Consistency of identities across cloud and on-premises will reduce human errors and resulting security risk.
Teams managing resources in both environments need a consistent authoritative source to achieve security
assurances.
For Azure, synchronize Azure AD with your existing authoritative on premises Active Directory Domain Services
(AD DS) using What is hybrid identity?.
This is also required for an Office 365 migration, so it is often already done before Azure migration and
development projects begin.
Note that administrator accounts should be excepted from synchronization as described in Don’t synchronize
on-premises admin accounts to cloud identity providers and Critical impact account dependencies.

Cloud provider identity source for third parties

Best practice: Use cloud identity services to host non-employee accounts such as vendors, partners, and
customers, rather than rather than including them in your on-premises directory.
This reduces risk by granting the appropriate level of access to external entities instead of the full default
permissions given to full-time employees. This least privilege approach and clear clearly differentiation of
external accounts from company staff makes it easier to prevent and detect attacks coming in from these
vectors. Additionally, management of these identities is done by the external also increases productivity by
parties, reducing effort required by company HR and IT teams.
For example, these capabilities natively integrate into the same Azure AD identity and permission model used by
Azure and Office 365:
Azure AD for employees and enterprise resources.
Azure AD B2B for business partners.
Azure AD B2C customers or citizens.
For more information, see the Azure AD federation compatibility list.

Passwordless or multi-factor authentication for administrative

accounts
Best practice: All users should be converted to use passwordless authentication or multi-factor authentication
(MFA) over time.
The details of this recommendation are in the administration section Passwordless Or multi-factor
authentication for admins
The same recommendation applies to all users but should be applied first and strongest to accounts with
administrative privileges.
You can also reduce use of passwords by applications using Managed Identities to grant access to resources in
Azure.

Block legacy authentication

Best practice: Disable insecure legacy protocols for internet-facing services.
Legacy authentication methods are among the top attack vectors for cloud-hosted services. Created before
multifactor-authentication existed, legacy protocols don’t support additional factors beyond passwords and are
therefore prime targets for password spraying, dictionary, or brute force attacks. As an example, nearly 100% of
all password spray attacks against Office 365 customers use legacy protocols. Additionally, these older protocols
frequently lack other attack countermeasures, such as account lockouts or back-off timers. Services running on
Microsoft’s cloud that block legacy protocols have observed a 66% reduction in successful account
compromises.
For Azure and other Azure AD-based accounts, configure Conditional Access to block legacy protocols.
Disabling legacy authentication can be difficult, as some users may not want to move to new client software that
supports modern authentication methods. However, moving away from legacy authentication can be done
gradually. Start by using metrics and logging from your authentication provider to determine the how many
users still authenticate with old clients. Next, disable any down-level protocols that aren’t in use, and set up
Conditional Access for all users who aren’t using legacy protocols. Finally, give plenty of notice and guidance to
users on how to upgrade before blocking legacy authentication for all users on all services at a protocol level.

No on-premises admin accounts in cloud identity providers

Best practice: Don’t synchronize accounts with the highest privilege access to on-premises resources as you
synchronize your enterprise identity systems with cloud directories.
This mitigates the risk of an adversary pivoting to full control of on-premises assets following a successful
compromise of a cloud account. This helps contain the scope of an incident from growing significantly.
For Azure, don’t synchronize accounts to Azure AD that have high privileges in your existing AD DS. This is
blocked by default in the default Azure AD Connect configuration, so you only need to confirm you haven’t
customized this configuration.
This is related to the critical impact account dependencies guidance in the administration section that mitigates
the inverse risk of pivoting from on-premises to cloud assets.

Modern password protection

Best practice: Provide modern and effective protections for accounts that cannot go passwordless
(Passwordless Or multi-factor authentication for admins).
Legacy identity providers mostly checked to make sure passwords had a good mix of character types and
minimum length, but we have learned that these controls in practice led to passwords with less entropy that
could be cracked easier:
Microsoft - https://www.microsoft.com/research/publication/password-guidance/
NIST - https://pages.nist.gov/800-63-3/sp800-63b.html
Identity solutions today need to be able to respond to types of attacks that didn't even exist one or two decades
ago such as password sprays, breach replays (also called “credential stuffing”) that test username/password
pairs from other sites’ breaches, and phishing man-in-the-middle attacks. Cloud identity providers are uniquely
positioned to offer protection against these attacks. Since they handle such large volumes of sign-ons, they can
apply better anomaly detection and use a variety of data sources to both proactively notify companies if their
users’ passwords have been found in other breaches, as well as validate that any given sign in appears
legitimate and is not coming from an unexpected or known-malicious host.
Additionally, synchronizing passwords to the cloud to support these checks also add resiliency during some
attacks. Customers affected by (Not)Petya attacks were able to continue business operations when password
hashes were synchronized to Azure AD (vs. near zero communications and IT services for customers affected
organizations that had not synchronized passwords).
For Azure, enable modern protections in Azure AD with these steps:
1. Configure Azure AD Connect to synchronize password hashes
2. Choose whether to automatically remediate these issues or manually remediate them based on a report:
a. Automatic Enforcement - Automatically remediate high risk passwords with Conditional Access
leveraging Azure AD Identity Protection risk assessments
b. Repor t & Manually Remediate - View reports and manually remediate accounts
Azure AD repor ting - Risk events are part of Azure AD's security reports. For more information,
see the users at risk security report and the risky sign-ins security report.
Azure AD Identity Protection - Risk events are also part of the reporting capabilities of Azure
Active Directory Identity Protection.
Use the Identity Protection risk events API to gain programmatic access to security detections using Microsoft
Graph.
Cross-platform credential management
Best practice: Use a single identity provider for authenticating all platforms (Windows, Linux, and others) and
cloud services.
A single identity provider for all enterprise assets will simplify management and security, minimizing the risk of
oversights or human mistakes. Deploying multiple identity solutions (or an incomplete solution) can result in
unenforceable password policies, passwords not reset after a breach, proliferation of passwords (often stored
insecurely), and former employees retaining passwords after termination.
For example, Azure AD can be used to authenticate:
Windows
Linux
Azure
Office 365
Amazon Web Services (AWS)
Google Services
Remote access to legacy on-premises applications
Third-party SaaS providers

Conditional Access for users with Zero Trust

Best practice: Authentication for all users should include measurement and enforcement of key security
attributes to support a Zero Trust strategy.
The details of this recommendation are in the administration section Enforce conditional access for ADMINS
(Zero Trust). The same recommendation applies to all users, but should be applied first to accounts with
administrative privileges.
You can also reduce use of passwords by applications using Managed Identities to grant access to resources in
Azure.

Simulate attacks
Best practice: Regularly simulate attacks against your users to educate and empower them.
People are a critical part of your defense, so ensure they have the knowledge and skills to avoid and resist
attacks will reduce your overall organizational risk.
You can use Attack Simulator in Microsoft Defender for Office 365 or any number of third-party offerings.

Next step
Review identity and device access capabilities.

See also
Zero Trust Security Model and Framework
Microsoft security documentation
 Identity and access management capabilities
12/13/2021 • 3 minutes to read • Edit Online

This article lists capabilities that can help with identity.

Cloud identity services

C A PA B IL IT Y DESC RIP T IO N SEE M O RE

Azure Active Directory (Azure AD) Establish a single Azure AD enterprise Azure Active Directory documentation
directory for managing identities of
full-time employees and enterprise
resources. A single authoritative source
increases clarity and consistency for all
roles in IT and security and reduces
security risk from human errors and
automation failures resulting from
complexity.

Azure AD Connect Synchronize Azure AD with your What is hybrid identity with Azure
existing authoritative on premises Active Directory?
Active Directory using Azure AD
connect. This is also required for an
Office 365 migration, so it is often
already done before Azure migration
and development projects begin.

Azure AD B2B collaboration Use Azure AD business-to-business Azure Active Directory B2B
(B2B) collaboration to host non- documentation
employee accounts like vendors and
partners. This reduces risk by granting
the appropriate level of access to
external entities instead of the full
default permissions given to full-time
employees. This least privilege
approach and clear clearly
differentiation of external accounts
from company staff makes it easier to
prevent and detect attacks coming in
from these vectors.

Azure AD B2C Use Azure AD B2C for consumer and Azure Active Directory B2C
citizen accounts. Azure AD B2C is an documentation
identity management service that
enables custom control of how your
customers sign up, sign in, and
manage their profiles when using your
iOS, Android, .NET, single-page (SPA),
and other applications.

Identity security capabilities

 C A PA B IL IT Y DESC RIP T IO N M O RE IN F O RM AT IO N

Azure Multi-Factor Authentication MFA provides additional security by How MFA works
(MFA) requiring a second form of
authentication.

Passwordless authentication Azure AD supports several methods of Passwordless authentication options

passwordless authentication, including for Azure Active Directory
Windows Hello for Business, Microsoft
Authenticator app, and FIDO2 security
keys. Passwordless authentication
methods are convenient because the
password is removed and replaced
with something you have, plus
something you are or something you
know.

Conditional Access With Conditional Access, Azure AD Conditional Access documentation

evaluates the conditions of the user
login and uses conditional access Common Conditional Access policies
policies you create to allow access. For
example, you can create a Conditional Recommended identity and device
Access policy to require device access configurations
compliance for access to sensitive data.
This greatly reduces the risk that a
person with a stolen identity can
access your sensitive data. It also
protects sensitive data on the devices,
because the devices must meet specific
requirements for health and security.

Azure AD self-service password reset SSPR allows your users to reset their How it works: Azure AD self-service
(SSPR) passwords securely and without password reset
helpdesk intervention, by providing
verification of multiple authentication
methods that the administrator can
control.

Azure Active Directory Identity Azure AD Identity Protection enables What is Azure Active Directory Identity
Protection you to detect potential vulnerabilities Protection?
affecting your organization's identities
and configure automated remediation
policy to low, medium, and high sign-
in risk and user risk.

Azure AD password protection for Protect on-premises Active Directory Enforce Azure AD password protection
Windows Server Active Directory accounts with Azure AD password for Windows Server Active Directory
protection. This does the same checks
on-premises as Azure AD does for
cloud-based changes. These checks are
performed during password changes
and password reset scenarios.

Additional identity security capabilities for administrators

 C A PA B IL IT Y DESC RIP T IO N M O RE IN F O RM AT IO N

Azure AD Privileged Identity Privileged Identity Management What is Azure AD Privileged Identity
Management provides time-based and approval- Management?
based role activation to mitigate the
risks of excessive, unnecessary, or
misused access permissions on
resources that you care about.

Managed Identities You can reduce use of passwords by What are managed identities for Azure
applications using Managed Identities resources?
to grant access to resources in Azure

See also
Zero Trust Security Model and Framework
Microsoft security documentation
 Microsoft Security Best Practices module: Network
security and containment
12/13/2021 • 2 minutes to read • Edit Online

Network Security & Containment helps reduce organizational risk by providing access controls to limit the
ability of attackers to traverse the enterprise environment without impeding legitimate communications and
interactions.
See the Network security and containment and Capabilities topics for more information.
The following videos provide guidance on network security and containment. You can also download the
PowerPoint slides associated with these videos.

Part 1: Introduction - Overview of Azure Network Security (21:31)

Part 2: Enterprise Consistency & Segmentation Alignment (04:15)

Part 3: Pragmatic Containment Strategy (04:14)

Part 4: Internet Edge Strategy (01:59)

Part 5: ExpressRoute Termination (02:24)

Part 6: Deprecating Legacy Technology (02:35)

Part 7: Subnet & NSG Design (03:04)

Part 8: DDoS Mitigations (02:41)

Part 9: Azure Ingress/Egress Security (02:08)

Part 10: Advanced Visibility (02:09)

Next steps
For additional security guidance from Microsoft, see Microsoft security documentation.
 Network security and containment
12/13/2021 • 11 minutes to read • Edit Online

Network security has been the traditional lynchpin of enterprise security efforts. However, cloud computing has
increased the requirement for network perimeters to be more porous and many attackers have mastered the art
of attacks on identity system elements (which nearly always bypass network controls). These factors have
increased the need to focus primarily on identity-based access controls to protect resources rather than
network-based access controls.
These do diminish the role of network security controls, but do not eliminate it entirely. While network security
is no longer the primary focus for securing cloud-based assets, it is still a top priority for the large portfolio of
legacy assets (which were built with the assumption that a network firewall-based perimeter was in place). Many
attackers still employ scanning and exploit methods across public cloud provider IP ranges, successfully
penetrating defenses for those who don’t follow basic network security hygiene. Network security controls also
provide a defense-in-depth element to your strategy that help protect, detect, contain, and eject attackers who
make their way into your cloud deployments.
In the category of network security and containment, we have the following best practice recommendations:
Align network segmentation with overall strategy
Centralize network management and security
Build a network containment strategy
Define an internet edge strategy

Centralize network management and security

Centralize the organizational responsibility for management and security of core networking functions such as
cross-premises links, virtual networking, subnetting, and IP address schemes as well as network security
elements such as virtual network appliances, encryption of cloud virtual network activity and cross-premises
traffic, network-based access controls, and other traditional network security components.
When you centralize network management and security you reduce the potential for inconsistent strategies that
can create potential attacker exploitable security risks. Because all divisions of the IT and development
organizations do not have the same level of network management and security knowledge and sophistication,
organizations benefit from leveraging a centralized network team’s expertise and tooling.
Microsoft Defender for Cloud can be used to help centralize the management of network security.

Align network segmentation with enterprise segmentation strategy

Align your network segmentation model with the enterprise segmentation model for your organization (defined
in Governance, Risk, and Compliance section).
This will reduce confusion and resulting challenges with different technical teams (networking, identity,
applications, etc.) each developing their own segmentation and delegation models that don’t align with each
other. This leads to a straightforward and unified security strategy, which helps reduce the number of errors due
to human error and inability to increase reliability through automation.
Please compare images in Network security and containment.
Evolve security beyond network controls
Ensure technical controls can effectively prevent, detect, and respond to threats outside the networks you
control.
As organizations shift to modern architectures, many services and components required for applications will be
accessed over the internet or on cloud provider networks, often by mobile and other devices off the network.
Traditional network controls based on a “trusted intranet” approach will not be able to effectively provide
security assurances for these applications. This shifting landscape is captured well by principles documented by
the Jericho Forum and ‘Zero Trust’ approaches.
Build a risk containment strategy based on a combination of network controls and application, identity, and
other control types.
Ensure that resource grouping and administrative privileges align to the segmentation model (see figure
XXXX)
Ensure you are designing security controls that identify and allow expected traffic, access requests, and
other application communications between segments. Monitor communications between segments to
identify on any unexpected communications so you can investigate whether to set alerts or block traffic
to mitigate risk of adversaries crossing segmentation boundaries.

Build a security containment strategy

Create a risk containment strategy that blends proven approaches including:
Existing network security controls and practices
Native security controls available in Azure
Zero-trust approaches
Containment of attack vectors within an environment is critical. However, in order to be effective in cloud
environments, traditional approaches may prove inadequate and security organizations need to evolve their
methods.
Consistency of controls across on-premises and cloud infrastructure is important, but defenses are more
effective and manageable when leveraging native capabilities provided by a cloud service provider,
 dynamic just-in-time (JIT) approaches, and integrated identity and password controls, such as those
recommended by zero trust/continuous validation approaches.
A network security best practice is to make sure there are network access controls between network
constructs. These constructs can represent virtual networks, or subnets within those virtual networks.
This works to protect and contain East-West traffic within your cloud network infrastructure.
An important network security design decision is to use or not use host-based firewalls. Host-based
firewalls support a comprehensive defense in-depth strategy. However, to be of most use they require
significant management overhead. If your organization has found them effective in helping you protect
and discover threats in the past, you might consider using them for your cloud-based assets. If you
discover that they have not added significant value, discontinue their use and explore native solutions on
your cloud service provider’s platform that deliver similar value.
An evolving emerging best practice recommendation is to adopt a Zero Trust strategy based on user, device, and
application identities. In contrast to network access controls that are based on elements such as source and
destination IP address, protocols, and port numbers, Zero Trust enforces and validates access control at “access
time”. This avoids the need to play a prediction game for an entire deployment, network, or subnet – only the
destination resource needs to provide the necessary access controls.
Azure Network Security Groups can be used for basic layer 3 & 4 access controls between Azure Virtual
Networks, their subnets, and the Internet.
Azure Web Application Firewall and the Azure Firewall can be used for more advanced network access
controls that require application layer support.
Local Admin Password Solution (LAPS) or a third-party Privileged Access Management can set strong
local admin passwords and just in time access to them
Additionally, third parties offer microsegmentation approaches that may enhance your network controls by
applying zero trust principles to networks you control with legacy assets on them.

Define an internet edge strategy

Choose whether to use native cloud service provider controls or virtual network appliances for internet edge
security.
Internet edge traffic (sometimes referred to as “North-South” traffic) represents network connectivity between
your assets in the cloud and the Internet. Legacy workloads require protection from Internet endpoints because
they were built with the assumption that an internet firewall protected them against these attacks. An Internet
edge strategy is intended to mitigate as many attacks from the internet as is reasonable to detect or block.
There are two primary choices that can provide Internet edge security controls and monitoring:
Cloud Service Provider Native Controls (Azure Firewall + Web Application Firewall (WAF))
Partner Virtual Network Appliances (Firewall and WAF Vendors available in Azure Marketplace)
Cloud service provider native controls typically offer basic security that is good enough for common
attacks, such as the OWASP Top 10.
In contrast, cloud service provider partner capabilities often provide much more advanced features that
can protect against sophisticated (but often uncommon) attacks. Partner solutions consistently cost more
than native controls. In addition, configuration of partner solutions can be very complex and more fragile
than native controls because they do not integrate with cloud’s fabric controllers.
The decision to use native versus partner controls should be based on your organization’s experience and
requirements. If the features of the advanced firewall solutions don’t provide sufficient return on investment,
you may consider using the native capabilities that are designed to be easy to configure and scale.

Discontinue legacy network security technology

Discontinue the use of signature-based Network Intrusion Detection/Network Intrusion Prevention (NIDS/NIPS)
Systems and Network Data Leakage/Loss Prevention (DLP).
The major cloud service providers already filter for malformed packets and common network layer attacks, so
there’s no need for a NIDS/NIPS solution to detect those. In addition, traditional NIDS/NIPS solutions are
typically driven by signature-based approaches (which are considered outdated) and are easily evaded by
attackers and typically produce a high rate of false positives.
Network-based DLP is decreasingly effective at identifying both inadvertent and deliberate data loss. The reason
for this is that most modern protocols and attackers use network-level encryption for inbound and outbound
communications. The only viable workaround for this is “SSL-bridging” which provides an “authorized man-in-
the-middle” that terminates and then reestablishes encrypted network connections. The SSL-bridging approach
has fallen out of favor because of the level of trust required for the partner running the solution and the
technologies that are being used.
Based on this rationale, we offer an all-up recommendation that you discontinue use of these legacy network
security technologies. However, if your organizational experience is that these technologies have had a palpable
impact on preventing and detecting real attacks, you can consider porting them to your cloud environment.

Design virtual network subnet security

Design virtual networks and subnets for growth.
Most organizations end up adding more resources to their networks than they initially planned for. When this
happens, IP addressing and subnetting schemes need to be refactored to accommodate the extra resources. This
is a labor-intensive process. There is limited security value in creating a very large number of small subnets and
then trying to map network access controls (such as security groups) to each of them.
We recommend that you plan your subnets based on common roles and functions that use common protocols
for those roles and functions. This allows you to add resources to the subnet without needing to make changes
to security groups that enforce network level access controls.
Do not use “all open” rules for inbound and outbound traffic to and from subnets. Use a network “least
privilege” approach and only allow relevant protocols. This will decrease your overall network attack surface on
the subnet.
All open rules (allowing inbound/outbound to and from 0.0.0.0-255.255.255.255) provide a false sense of
security since such a rule enforces no security at all.
However, the exception to this is if you want to use security groups only for network logging. We do not
recommend this, but it is an option if you have another network access control solution in place.
Azure Virtual Network subnets can be designed in this way.

Mitigate DDoS attacks

Enable Distributed Denial of Service (DDoS) mitigations for all business-critical web application and services.
DDoS attacks are prevalent and are easily carried out by unsophisticated attackers. There are even “DDoS as a
service” options on the dark net. DDoS attacks can be very debilitating and completely block access to your
services and even take down the services, depending on the type of DDoS attack.
The major cloud service providers offer DDoS protection of services of varying effectiveness and capacity. The
cloud service providers typically provide two DDoS protection options:
DDoS protection at the cloud network fabric level – all customers of the cloud service provider benefit
from these protections. The protection is usually focused at the network (layer 3) level.
DDoS protection at higher levels that profile your services – this kind of protection will baseline your
deployments and then use machine learning techniques to detect anomalous traffic and proactively
protect based on their protection before there is service degradation
We recommend that you adopt the advance protection for any services where downtime will have negative
impact on the business.
An example of advanced DDoS protection is the Azure DDoS Protection Service.

Decide upon an internet ingress/egress policy

Choose to route traffic destined for the Internet through on-premises security devices or allow Internet
connectivity through cloud-based network security devices.
Routing Internet traffic through on-premises network security devices is also known as “forced tunneling”. In a
forced tunneling scenario, all connectivity to the Internet is forced back to on-premises network security devices
through a cross-premises WAN link. The goal is to provide network security teams greater security and visibility
for Internet traffic. Even when your resources in the cloud try to respond to incoming requests from the Internet,
the responses will be forced through on-premises network security devices.
Alternately, forced tunneling fits a “datacenter expansion” paradigm and can work well for a quick proof of
concept, but scales poorly because of the increased traffic load, latency, and cost.
The recommended approach for production enterprise use is to allow cloud resources to initiate and respond to
Internet request directly through cloud network security devices defined by your internet edge strategy.
The direct Internet approach fits the Nth datacenter paradigm (for example, Azure datacenters are a natural part
of my enterprise). This approach scales much better for an enterprise deployment as it removes hops that add
load, latency, and cost.
We recommend that you avoid forced tunneling for the reasons noted above.

Enable enhanced network visibility

You should enable enhanced network visibility by integrating network logs into a Security information and event
management (SIEM) like Microsoft Sentinel or a third partner solution such as Splunk, QRadar, or ArcSight ESM.
Integrating logs from your network devices, and even raw network traffic itself, will provide you greater visibility
over potential security threats flowing over the wire. This log information can be integrated in advanced SIEM
solutions or other analytics platforms. The modern machine learning based analytics platforms support
ingestion of extremely large amounts of information and can analyze large datasets very quickly. In addition,
these solutions can be tuned to significantly reduce false positive alerts.
Examples of network logs that provide visibility include:
Security group logs – flow logs and diagnostic logs
Web application firewall logs
Virtual network taps and their equivalents
Azure Network Watcher
Next steps
For additional security guidance from Microsoft, see Microsoft security documentation.
 Network security and containment capabilities
12/13/2021 • 3 minutes to read • Edit Online

This article lists capabilities that can help with network traffic and containment.

Capabilities that work with SaaS, PaaS, and IaaS apps

C A PA B IL IT Y DESC RIP T IO N M O RE IN F O RM AT IO N

Azure ExpressRoute Use Azure ExpressRoute to create Azure ExpressRoute

private connections between Azure
datacenters and infrastructure on your Azure ExpressRoute for Office 365
premises or in a colocation
environment. ExpressRoute
connections don't go over the public
Internet, and they offer more reliability,
faster speeds, and lower latencies than
typical Internet connections.

Capabilities that work with PaaS and IaaS apps

C A PA B IL IT Y DESC RIP T IO N M O RE IN F O RM AT IO N

Azure Application Gateway Azure Application Gateway is a web What is Azure Application Gateway?
traffic load balancer that enables you
to manage traffic to your web
applications. Traditional load balancers
operate at the transport layer (OSI
layer 4 - TCP and UDP) and route
traffic based on source IP address and
port, to a destination IP address and
port. With Application Gateway, you
can make routing decisions based on
additional attributes of an HTTP
request, such as URI path or host
headers.

Azure Traffic Manager Azure Traffic Manager is a DNS-based What is Traffic Manager?
traffic load balancer that enables you
to distribute traffic optimally to
services across global Azure regions,
while providing high availability and
responsiveness. Traffic Manager uses
DNS to direct client requests to the
most appropriate service endpoint
based on a traffic-routing method and
the health of the endpoints.

Additional capabilities for IaaS apps

C A PA B IL IT Y DESC RIP T IO N M O RE IN F O RM AT IO N

Azure Virtual Network Azure Virtual Network (VNet) is the What is Azure Virtual Network?
fundamental building block for your
private network in Azure. VNet enables Virtual Network documentation
many types of Azure resources, such
as Azure Virtual Machines (VM), to
securely communicate with each other,
the internet, and on-premises
networks. VNet is similar to a
traditional network that you'd operate
in your own data center, but brings
with it additional benefits of Azure's
infrastructure such as scale, availability,
and isolation.

When you create a VNet, you

configure the following elements:
address space, subnets, regions, and
subscription.

Point-to-site virtual private network You can connect your on-premises Point-to-site VPN
(VPN) and Site-to-site VPN computers and networks to a virtual
network using any combination of Site-to-site VPN
these VPN options and Azure
ExpressRoute.

Security groups and Network virtual You can filter network traffic between Network security groups
appliances subnets using either or both of these
options. Application security groups

Network security groups and Network virtual appliances (Azure

application security groups can contain Marketplace)
multiple inbound and outbound
security rules that enable you to filter
traffic to and from resources by source
and destination IP address, port, and
protocol.

A network virtual appliance is a VM

that performs a network function, such
as a firewall, WAN optimization, or
other network function.

Route tables and border gateway Azure routes traffic between subnets, Route tables
protocol (BGP) routes connected virtual networks, on-
premises networks, and the Internet, About BGP with Azure VPN Gateway
by default. You can implement either
or both of the options to override the
default routes Azure creates.
 C A PA B IL IT Y DESC RIP T IO N M O RE IN F O RM AT IO N

Azure DDoS Protection Azure DDoS protection, combined with Azure DDoS Protection Standard
application design best practices, overview
provide defense against DDoS attacks.

Azure DDoS Protection Basic is

automatically enabled as part of the
Azure platform and provides real-time
mitigation of common network-level
attacks.

DDoS Protection Standard provides

additional mitigation capabilities that
are tuned specifically to Azure Virtual
Network resources. DDoS Protection
Standard can mitigate the following
types of attacks: volumetric attacks,
protocol attacks, and resource
(application) layer attacks.

Azure Firewall Cloud-native network security to Azure Firewall

protect your Azure Virtual Network
resources.

Next steps
For additional security guidance from Microsoft, see Microsoft security documentation.
 Securing privileged access
12/13/2021 • 2 minutes to read • Edit Online

Organization's should make securing privileged access the top security priority because of the significant
potential business impact (and high likelihood) of attackers compromising this level of access.
Privileged access includes IT administrators with control of large portions of the enterprise estate and other
users with access to business critical assets.
Attackers frequently exploit weaknesses in privileged access security during human operated ransomware
attacks and targeted data theft. Privileged access accounts and workstations are so attractive to attackers
because these targets allow them to rapidly gain broad access to the business assets in the enterprise, often
resulting in rapid and significant business impact.
The following diagram summarizes the recommended privileged access strategy to create an isolated virtual
zone that these sensitive accounts can operate in with low risk.

Securing privileged access effectively seals off unauthorized pathways completely and leaves a select few
authorized access pathways that are protected and closely monitored. This diagram is discussed in more detail
in the article, Privileged Access Strategy.
Building this strategy requires a holistic approach combining multiple technologies to protect and monitor those
authorized escalation paths using Zero Trust principles including explicit validation, least privilege, and assume
breach. This strategy requires multiple complementary initiatives that establish a holistic technology approach,
clear processes, and rigorous operational execution to build and sustain assurances over time.

Get started and measure progress

Rapid Modernization Plan Best practices

(RaMP) Videos and Slides
- Plan and implement the
most impactful quick wins
Industry references
Securing privileged access is also addresses by these industry standards and best practices.

UK National Cyber Security Center Australian Cyber Security Center MITRE ATT&CK
(NCSC) (ACSC)

Next steps
Strategy, design, and implementation resources to help you rapidly secure privileged access for your
environment.

A RT IC L E DESC RIP T IO N

Strategy Overview of privileged access strategy

Success criteria Strategic success criteria

Security levels Overview of security levels for

accounts, devices, intermediaries, and
interfaces

Accounts Guidance on security levels and

controls for accounts

Intermediaries Guidance on security levels and

controls for intermediaries

Interfaces Guidance on security levels and

controls for interfaces
A RT IC L E DESC RIP T IO N

Devices Guidance on security levels and

controls for devices and workstations

Enterprise access model Overview of Enterprise Access Model

(successor to legacy tier model)

ESAE Retirement Information on retirement of legacy

administrative forest
 Privileged access: Strategy
12/13/2021 • 10 minutes to read • Edit Online

Microsoft recommends adopting this privileged access strategy to rapidly lower the risks to your organization
from high impact and high likelihood attacks on privileged access.
Privileged access should be the top security priority at ever y organization. Any compromise of these
users has a high likelihood of significant negative impact to the organization. Privileged users have access to
business critical assets in an organization, nearly always causing major impact when attackers compromise their
accounts.
This strategy is built on Zero Trust principles of explicit validation, least privilege, and assumption of breach.
Microsoft has provided implementation guidance to help you rapidly deploy protections based on this strategy

IMPORTANT
There is no single "silver bullet" technical solution that will magically mitigate privileged access risk, you must blend
multiple technologies together into a holistic solution that protects against multiple attacker entry points. Organizations
must bring the right tools for each part of the job.

Why is privileged access important?

Security of privileged access is critically important because it is foundational to all other security assurances, an
attacker in control of your privileged accounts can undermine all other security assurances. From a risk
perspective, loss of privileged access is a high impact event with a high likelihood of happening that is growing
at an alarming rate across industries.
These attack techniques were initially used in targeted data theft attacks that resulted in many high profile
breaches at familliar brands (and many unreported incidents). More recently these techniques have been
adopted by ransomware attackers, fueling an explosive growth of highly profitable human operated
ransomware attacks that intentionally disrupt business operations across industry.

IMPORTANT
Human operated ransomware is different from commodity single computer ransowmare attacks that target a single
workstation or device.

This graphic describes how this extortion based attack is growing in impact and likelihood using privileged
access:
 High business impact
It is difficult to overstate the potential business impact and damage of a loss to privileged access.
Attacker's with privileged access effectively have full control of all enterprise assets and resources,
giving them the ability to disclose any confidential data, stop all business processes, or subvert
business processes and machines to damage property, hurt people, or worse. Massive business
impact has been seen across every industry with:
Targeted data theft - attackers use privileged access to access and steal sensitive intellectual
property for their own use it or to sell/transfer to your competitors or foreign governments
Human-operated ransomware (HumOR) - attackers use privileged access to steal and/or
encrypt all data and systems in the enterprise, often stopping all business operations. They then
extort the target organization by demanding money to not disclose the data and/or providing
the keys to unlock it.
High likelihood of occurrence
The prevalence of privileged access attacks has grown since the advent of modern credential theft
attacks starting with pass the hash techniques. These techniques first jumped in popularity with
criminals starting with the 2008 release of the attack tool "Pass-the-Hash Toolkit" and have grown into
a suite of reliable attack techniques (mostly based on the Mimikatz toolkit). This weaponization and
automation of techniques allowed the attacks (and their subsequent impact) to grow at a rapid rate,
limited only by the target organization's vulnerability to the attacks and the attacker's
monetization/incentive models.
Prior to the advent of human-operated ransomware (HumOR), these attacks were prevalent but
often unseen or misunderstood because of:
Attacker monetization limits - Only groups and individuals who knew how to
monetize sensitive intellectual property from target organizations could profit from
these attacks.
Silent impact - Organizations often missed these attacks because they didn't have
detection tools, and also had a hard time seeing and estimating the resulting business
impact (for example, how their competitors were using their stolen intellectual property
and how that affected prices and markets, sometimes years later). Additionally,
organizations who saw the attacks often stayed silent about them to protect their
reputations.
Both the silent impact and attacker monetization limitations on these attacks are disintegrating
with the advent of human operated ransomware, which is growing in volume, impact, and
awareness because it is both:
Loud and disruptive - to business processes to payment of extortion demands.
Universally applicable - Every organization in every industry is financially motivated
to continue operations uninterrupted.
For these reasons, privileged access should be the top security priority at every organization.
Building your privileged access strategy
Privileged access strategy is a journey that must be composed of quick wins and incremental progress. Each
step in your privileged access strategy must take you closer to "seal" out persistent and flexible attackers from
privileged access, who are like water trying to seep into your environment through any available weakness.
This guidance is designed for all enterprise organizations regardless of where you already are in the journey.
Holistic practical strategy
Reducing risk from privileged access requires a thoughtful, holistic, and prioritized combination of risk
mitigations spanning multiple technologies.
Building this strategy requires recognition that attackers are like water as they have numerous options they can
exploit (some of which can appear insignificant at first), attackers are flexible in which ones they use, and they
generally take the path of least resistance to achieving their objectives.

The paths attackers prioritize in actual practice are a combination of:

Established techniques (often automated into attack tools)
New techniques that are easier to exploit
Because of the diversity of technology involved, this strategy requires a complete strategy that combines
multiple technologies and follows Zero Trust principles.

IMPORTANT
You must adopt a strategy that includes multiple technologies to defend against these attacks. Simply implementing a
prvileged identity management / privileged access management (PIM/PAM) solution is not sufficient. For more
information see, Privileged access Intermediaries.

The attackers are goal-oriented and technology agnostic, using any type of attack that works.
The access control backbone you are defending is integrated into most or all systems in the enterprise
environment.
Expecting you can detect or prevent these threats with just network controls or a single privileged access
solution will leave you vulnerable to many other types of attacks.
Strategic assumption - Cloud is a source of security
This strategy uses cloud services as the primary source of security and management capabilities rather than on-
premises isolation techniques for several reasons:
Cloud has better capabilities - The most powerful security and management capabilities available today
come from cloud services, including sophisticated tooling, native integration, and massive amounts of
security intelligence like the 8+ trillion security signals a day Microsoft uses for our security tools.
Cloud is easier and faster - Adopting cloud services requires little to no infrastructure for implementing
and scaling up, enabling your teams to focus on their security mission rather than technology integration.
Cloud requires less maintenance - The cloud is also managed, maintained, and secured consistently by
vendor organizations with teams dedicated to that single purpose for thousands of customer organizations,
reducing the time and effort for your team to rigorously maintain capabilities.
Cloud keeps improving - Features and functionality in cloud services are constantly being updated
without a need for your organization to invest ongoing.
Building the recommended strategy
Microsoft's recommended strategy is to incrementally build a 'closed loop' system for privileged access that
ensures only trustworthy 'clean' devices, accounts, and intermediary systems can be used for privileged access
to business sensitive systems.
Much like waterproofing something complex in real life like a boat, you need to design this strategy with an
intentional outcome, establish and follow standards carefully, and continually monitor and audit the outcomes
so that you remediate any leaks. You wouldn't just nail boards together in a boat shape and magically expect a
waterproof boat. You would focus first on building and waterproofing significant items like the hull and critical
components like the engine and steering mechanism (while leaving ways for people to get in), then later
waterproofing comfort items like radios, seats, and the like. You would also maintain it over time as even the
most perfect system could spring a leak later, so you need to keep up with preventive maintenance, monitor for
leaks, and fix them to keep it from sinking.
Securing Privileged Access has two simple goals
1. Strictly limit the ability to perform privileged actions to a few authorized pathways
2. Protect and closely monitor those pathways
There are two types of pathways to accessing the systems, user access (to use the capability) and privileged
access (to manage the capability or access a sensitive capability)

User Access - the lighter blue path on the bottom of the diagram depicts a standard user account performing
general productivity tasks like email, collaboration, web browsing, and use of line-of-business applications or
websites. This path includes an account logging on to a device or workstations, sometimes passing through
an intermediary like a remote access solution, and interacting with enterprise systems.
Privileged Access - the darker blue path on the top of the diagram depicts privileged access, where privileged
accounts like IT Administrators or other sensitive accounts access business critical systems and data or
perform administrative tasks on enterprise systems. While the technical components may be similar in
 nature, the damage an adversary can inflict with privileged access is much higher.
The full access management system also includes identity systems and authorized elevation paths.

Identity Systems - provide identity directories that host the accounts and administrative groups,
synchronization and federation capabilities, and other identity support functions for standard and privileged
users.
Authorized Elevation Paths - provide means for standard users to interact with privileged workflows, such as
managers or peers approving requests for administrative rights to a sensitive system through a just in time
(JIT) process in a Privileged Access Management / Privileged Identity management system.
These components collectively comprise the privileged access attack surface that an adversary may target to
attempt gaining elevated access to your enterprise:

NOTE
For on-premises and infrastructure as a service (IaaS) systems hosted on a customer managed operating system, the
attack surface dramatically increases with management and security agents, service accounts, and potential configuration
issues.

Creating a sustainable and manageable privileged access strategy requires closing off all unauthorized vectors
to create the virtual equivalent of a control console physically attached to a secure system that represents the
only way to access it.
This strategy requires a combination of:
Zero Trust access control described throughout this guidance, including the rapid modernization plan
(RAMP)
Asset protection to protect against direct asset attacks by applying good security hygiene practices to these
 systems. Asset protection for resources (beyond access control components) is out of scope of this guidance,
but typically includes rapid application of security updates/patches, configuring operating systems using
manufacturer/industry security baselines, protecting data at rest and in transit, and integrating security best
practices to development / DevOps processes.

Strategic initiatives in the journey

Implementing this strategy requires four complementary initiatives that each have clear outcomes and success
criteria
1. End-to-end Session Security - Establish explicit Zero Trust validation for privileged sessions, user sessions,
and authorized elevation paths.
a. Success Criteria: Each session will validate that each user accounts and device are trusted at a
sufficient level before allowing access.
2. Protect & Monitor Identity Systems including Directories, Identity Management, Admin Accounts, Consent
grants, and more
a. Success Criteria: Each of these systems will be protected at a level appropriate for the potential
business impact of accounts hosted in it.
3. Mitigate Lateral Traversal to protect against lateral traversal with local account passwords, service account
passwords, or other secrets
a. Success Criteria: Compromising a single device will not immediately lead to control of many or all
other devices in the environment
4. Rapid Threat Response to limit adversary access and time in the environment
a. Success Criteria: Incident response processes impede adversaries from reliably conducting a multi-
stage attack in the environment that would result in loss of privileged access. (as measured by
reducing the mean time to remediate (MTTR) of incidents involving privileged access to near zero and
reducing MTTR of all incidents to a few minutes so adversaries don't have time to target privileged
access)

Next steps
Securing privileged access overview
Measuring success
Security levels
Privileged access accounts
Intermediaries
Interfaces
Privileged access devices
Enterprise access model
Enhanced Security Admin Environment (ESAE) retirement
 Success criteria for privileged access strategy
12/13/2021 • 7 minutes to read • Edit Online

This document describes the success criteria for a privileged access strategy. This section describes strategic
perspectives of success for a privileged access strategy. For a roadmap on how to adopt this strategy, see the
rapid modernization plan (RaMP). For implementation guidance, see privileged access deployment
Implementing a holistic strategy using Zero Trust approaches creates a "seal" of sorts over the access control for
privileged access that makes it resistant to attackers. This strategy is accomplished by limiting pathways to
privileged access only a select few, and then closely protecting and monitoring those authorized pathways.

A successful strategy must address the all points attackers can use to intercept privileged access workflows
including four distinct initiatives:
Privileged Access workflow elements of the privileged access workflow including underlying devices,
operating systems, applications, and identities
Identity systems hosting the privileged accounts and the groups, and other artifacts that confer privilege
on the accounts
User access workflow and authorized elevation paths that can lead to privileged access
Application interfaces where zero trust access policy is enforced and role-based access control (RBAC) is
configured to grant privileges

NOTE
A complete security strategy also includes asset protections that are beyond the scope of access control, such as data
backups and protections against attacks on the application itself, the underlying operating system and hardware, on
service accounts used by the application or service, and on data while at rest or in transit. For more information on
modernizing a security strategy for cloud, see the article Define a security strategy.

An attack consists of human attackers leveraging automation and scripts to attack an organization is composed
of humans, the processes they follow, and the technology they use. Because of this complexity of both attackers
and defenders, the strategy must be multi-faceted to guard against all the people, process, and technology ways
that the security assurances could inadvertently be undermined.
Ensuring sustainable long-term success requires meeting the following criteria:
Ruthless prioritization
Balance security and productivity
Strong partnerships within the organization
Disrupt attacker return on investment
Follow clean source principle

Ruthless prioritization
Ruthless prioritization is the practice of taking the most effective actions with the fastest time to value first, even
if those efforts don't fit pre-existing plans, perceptions, and habits. This strategy lays out the set of steps that
have been learned in the fiery crucible of many major cybersecurity incidents. The learnings from these
incidents form the steps we help organizations take to ensure that these crises don't happen again.
While it's always tempting for security professionals to try to optimize familiar existing controls like network
security and firewalls for newer attacks, this path consistently leads to failure. Microsoft's Detection and
Response Team (DART) has been responding to privileged access attacks for nearly a decade and consistently
sees these classic security approaches fail to detect or stop these attacks. While network security provides
necessary and important basic security hygiene, it's critical to break out of these habits and focus on mitigations
that will deter or block real world attacks.
Ruthlessly prioritize the security controls recommended in this strategy, even if it challenges existing
assumptions and forces people to learn new skills.

Balance security and productivity

As with all elements of security strategy, privileged access should ensure that both productivity and security
goals are met.
Balancing security avoids the extremes that create risk for the organization by:
Avoiding overly strict security that causes users to go outside the secure policies, pathways, and systems.
Avoiding weak security that harms productivity by allowing adversaries to easily compromise the
organization.
For more information about security strategy, see the article Defining a security strategy.
To minimize negative business impact from security controls, you should prioritize invisible security controls
that improve user workflows, or at least don't impede or change user workflows. While security sensitive roles
may need visible security measures that change their daily workflows to provide security assurances, this
implementation should be done thoughtfully to limit the usability impact and scope as much as possible.
This strategy follows this guidance by defining three profiles (detailed later in Keep it Simple - Personas and
Profiles)
Strong partnerships within the organization
Security must work to build partnerships within the organization to be successful. In addition to the timeless
truth that "none of us is as smart as all of us," the nature of security is to be a support function to protect
someone else's resources. Security isn't accountable for the resources they help protect (profitability, uptime,
performance, etc.), security is a support function that provides expert advice and services to help protect the
intellectual property and business functionality that is important to the organization.
Security should always work as a par tner in support of business and mission objectives. While security
should not shy away from giving direct advice like recommending against accepting a high risk, security should
also always frame that advice in terms of the business risk relative to other risks and opportunities managed by
the resource owners.
While some parts of security can be planned and executed successfully mostly within security organization,
many like securing privileged access require working closely with IT and business organizations to understand
which roles to protect, and help update and redesign workflows to ensure they are both secure and allow people
to do their jobs. For more information on this idea, see the section Transformations, mindsets, and expectations
in the security strategy guidance article.

Disrupt attacker return on investment

Maintain focus on pragmatism by ensuring that defensive measures are likely to meaningfully disrupt the
attacker value proposition of attacking you, increasing cost and friction on the attacker's ability to successfully
attack you. Evaluating how defensive measures would impact the adversary's cost of attack provides both a
healthy reminder to focus on the attackers perspective as well as a structured mechanism to compare the
effectiveness of different mitigation options.
Your goal should be to increase the attackers cost while minimizing your own security investment level:
Disrupt attacker return on investment (ROI) by increasing their cost of attack across the elements of the
privileged access session. This concept is described in more detail in the article Success criteria for privileged
access strategy.

IMPORTANT
A privileged access strategy should be comprehensive and provide defense in depth, but must avoid the Expense in depth
fallacy where defenders simply pile on more same (familiar) type controls (often network firewalls/filters) past the point
where they add any meaningful security value.

For more information on attacker ROI, see the short video and in-depth discussion Disrupting attacker return on
investment.

Clean source principle

The clean source principle requires all security dependencies to be as trustworthy as the object being secured.

Any subject in control of an object is a security dependency of that object. If an adversary can control anything
in control of a target object, they can control that target object. Because of this threat, you must ensure that the
assurances for all security dependencies are at or above the desired security level of the object itself. This
principle applies across many types of control relationships:
While simple in principle, this concept gets complex easily in the real world as most enterprises grew organically
over decades and have many thousands of control relationships recursively that build on each other, loop back
on each other, or both. This web of control relationships provides many access paths that an attacker can
discover and navigate during an attack, often with automated tools.
Microsoft's recommended privileged access strategy is effectively a plan to untangle the most important parts
of this knot first using a Zero Trust approach, by explicitly validating that the source is clean before allowing
access to the destination.
In all cases, the trust level of the source must be the same or higher than the destination.
The only notable exception to this principle is allowing the use of unmanaged personal devices and partner
devices for enterprise scenarios. This exception enables enterprise collaboration and flexibility and can be
mitigated to an acceptable level for most organizations because of the low relative value of the enterprise
assets. For more context on BYOD security, see the blog post How a BYOD policy can reduce security risk in
the public sector.
This same exception cannot be extended to specialized security and privileged security levels however
because of the security sensitivity of these assets. Some PIM/PAM vendors may advocate that their solutions
can mitigate device risk from lower-level devices, but we respectfully disagree with those assertions based
on our experience investigating incidents. The asset owners in your organization may choose to accept risk of
using enterprise security level devices to access specialized or privileged resources, but Microsoft does not
recommend this configuration. For more information, see the intermediary guidance for Privileged Access
Management / Privileged Identity management.
The privileged access strategy accomplishes this principle primarily by enforcing Zero Trust policy with
Conditional Access on inbound sessions at interfaces and intermediaries. The clean source principle starts with
getting a new device from an OEM that is built to your security specifications including operating system
version, security baseline configuration, and other requirements such as using Windows Autopilot for
deployment.
Optionally, the clean source principle can extend into a highly rigorous review of each component in the supply
chain including installation media for operating systems and applications. While this principle would be
appropriate for organizations facing highly sophisticated attackers, it should be a lower priority than the other
controls in this guidance.

Next steps
Securing privileged access overview
Privileged access strategy
Security levels
Privileged access accounts
Intermediaries
Interfaces
Privileged access devices
Enterprise access model
 Privileged access security levels
12/13/2021 • 4 minutes to read • Edit Online

This document describes the security levels of a privileged access strategy For a roadmap on how to adopt this
strategy, see the rapid modernization plan (RaMP). For implementation guidance, see privileged access
deployment
These levels are primarily designed to provide simple and straightforward technical guidance so that
organizations can rapidly deploy these critically important protections. The privileged access strategy recognizes
that organizations have unique needs, but also that custom solutions create complexity that results in higher
costs and lower security over time. To balance this need, the strategy provides firm prescriptive guidance for
each level and flexibility through allowing organizations to choose when each role will be required to meet the
requirements of that level.

Making things simple helps people understand it and lowers the risk they will be confused and make mistakes.
While the underlying technology is almost always complex, it is critical to keep things simple rather than
creating custom solutions that are difficult to support. For more information, see the article Security design
principles.
Designing solutions that are focused on the needs of the administrators and end users, will keep it simple for
them. Designing solutions that are simple for security and IT personnel to build, assess, and maintain (with
automation where possible) leads to less security mistakes and more reliable security assurances.
The recommended privileged access security strategy implements a simple three level system of assurances,
that span across areas, designed to be easy to deploy for: accounts, devices, intermediaries, and interfaces.
Each successive level drives up attacker costs, with additional level of Defender for Cloud investment. The levels
are designed to target the 'sweet spots' where defenders get the most return (attacker cost increase) for each
security investment they make.
Each role in your environment should be mapped to one of these levels (and optionally increased over time as
part of a security improvement plan). Each profile is clearly defined as a technical configuration and automated
where possible to ease deployment and speed up security protections. For implementation details see the
article, Privileged access roadmap.
The security levels used throughout this strategy are:
Enterprise security is suitable for all enterprise users and productivity scenarios. In the progression of
the rapid modernization plan, enterprise also serves as the starting point for specialized and privileged
access as they progressively build on the security controls in enterprise security.

NOTE
Weaker security configurations do exist, but aren't recommended by Microsoft for enterprise organizations today
because of the skills and resources attackers have available. For information on what attackers can buy from each
other on the dark markets and average prices, see the video Top 10 Best Practices for Azure Security

Specialized security provides increased security controls for roles with an elevated business impact (if
compromised by an attacker or malicious insider).
Your organization should have documented criteria for specialized and privileged accounts (for example,
potential business impact is over $1M USD) and then identify all the roles and accounts meeting that
criteria. (used throughout this strategy, including in the Specialized Accounts)
Specialized roles typically include:
Developers of business critical systems.
Sensitive business roles such as users of SWIFT terminals, researchers with access to sensitive
data, personnel with access to financial reporting prior to public release, payroll administrators,
approvers for sensitive business processes, and other high impact roles.
Executives and personal assistants / administrative assistants that that regularly handle sensitive
information.
High impact social media accounts that could damage the company reputation.
Sensitive IT Admins with a significant privileges and impact, but are not enterprise-wide. This group
typically includes administrators of individual high impact workloads. (for example, enterprise
resource planning administrators, banking administrators, help desk /tech support roles, etc.)
Specialized Account security also serves as an interim step for privileged security, which further builds on
these controls. See privileged access roadmap for details on recommended order of progression.
Privileged security is the highest level of security designed for roles that could easily cause a major
incident and potential material damage to the organization in the hands of an attacker or malicious
insider. This level typically includes technical roles with administrative permissions on most or all
enterprise systems (and sometimes includes a select few business critical roles)
Privileged accounts are focused on security first, with productivity defined as the ability to easily and
securely perform sensitive job tasks securely. These roles will not have the ability to do both sensitive
work and general productivity tasks (browse the web, install and use any app) using the same account or
the same device/workstation. They will have highly restricted accounts and workstations with increased
monitoring of their actions for anomalous activity that could represent attacker activity.
Privileged access security roles typically include:
 Azure AD Global Administrators and related roles
Other identity management roles with administrative rights to an enterprise directory, identity
synchronization systems, federation solution, virtual directory, privileged identity/access management
system, or similar.
Roles with membership in these on-premises Active Directory groups
Enterprise Admins
Domain Admins
Schema Admin
BUILTIN\Administrators
Account Operators
Backup Operators
Print Operators
Server Operators
Domain Controllers
Read-only Domain Controllers
Group Policy Creator Owners
Cryptographic Operators
Distributed COM Users
Sensitive on-premises Exchange groups (including Exchange Windows Permissions and
Exchange Trusted Subsystem)
Other Delegated Groups - Custom groups that may be created by your organization to manage
directory operations.
Any local administrator for an underlying operating system or cloud service tenant that is
hosting the above capabilities including
Members of local administrators group
Personnel who know the root or built in administrator password
Administrators of any management or security tool with agents installed on those
systems

Next steps
Securing privileged access overview
Privileged access strategy
Measuring success
Privileged access accounts
Intermediaries
Interfaces
Privileged access devices
Enterprise access model
 Privileged access: Accounts
12/13/2021 • 5 minutes to read • Edit Online

Account security is a critical component of securing privileged access. End to end Zero Trust security for sessions
requires strongly establishing that the account being used in the session is actually under the control of the
human owner and not an attacker impersonating them.
Strong account security starts with secure provisioning and full lifecycle management through to
deprovisioning, and each session must establish strong assurances that the account isn't currently compromised
based on all available data including historical behavior patterns, available threat intelligence, and usage in the
current session.

Account security
This guidance defines three security levels for account security that you can use for assets with different
sensitivity levels:

These levels establish clear and implementable security profiles appropriate for each sensitivity level that you
can assign roles to and scale out rapidly. All of these account security levels are designed to maintain or improve
productivity for people by limiting or eliminating interruption to user and admin workflows.

Planning account security

This guidance outlines the technical controls required to meet each level. Implementation guidance is in the
privileged access roadmap.
Account security controls
Achieving security for the interfaces requires a combination of technical controls that both protect the accounts
and provide signals to be used in a Zero Trust policy decision (see Securing Interfaces for policy configuration
reference).
The controls used in these profiles include:
Multi-factor authentication - providing diverse sources of proof that the (designed to be as easy as possible
for users, but difficult for an adversary to mimic).
Account risk - Threat and Anomaly Monitoring - using UEBA and Threat intelligence to identify risky
 scenarios
Custom monitoring - For more sensitive accounts, explicitly defining allowed/accepted behaviors/patterns
allows early detection of anomalous activity. This control is not suitable for general purpose accounts in
enterprise as these accounts need flexibility for their roles.
The combination of controls also enables you to improve both security and usability - for example a user who
stays within their normal pattern (using the same device in same location day after day) does not need to be
prompted for outside MFA every time they authenticate.

Enterprise security accounts

The security controls for enterprise accounts are designed to create a secure baseline for all users and provide a
secure foundation for specialized and privileged security:
Enforce strong multi-factor authentication (MFA) - Ensure that the user is authenticated with strong MFA
provided by an enterprise-managed identity system (detailed in the diagram below). For more
information about multi-factor authentication, see Azure security best practice 6.

NOTE
While your organization may choose to use an existing weaker form of MFA during a transition period, attackers
are increasingly evading the weaker MFA protections, so all new investment into MFA should be on the strongest
forms.

Enforce account/session risk - ensure that the account is not able to authenticate unless it is at a low (or
medium?) risk level. See Interface Security Levels for details on conditional enterprise account security.
Monitor and respond to alerts - Security operations should integrate account security alerts and get
sufficient training on how these protocols and systems work to ensure they are able to rapidly
comprehend what an alert means and react accordingly.
Enable Azure AD Identity Protection
Investigate risk Azure AD Identity Protection
Troubleshoot/Investigate Conditional Access Sign-in failures
The following diagram provides a comparison to different forms of MFA and passwordless authentication. Each
option in the best box is considered both high security and high usability. Each has different hardware
requirements so you may want to mix and match which ones apply to different roles or individuals. All Microsoft
passwordless solutions are recognized by Conditional Access as multi-factor authentication because they
require combining something you have with either biometrics, something you know, or both.
 NOTE
For more information on why SMS and other phone based authentication is limited, see the blog post It's Time to Hang
Up on Phone Transports for Authentication.

Specialized accounts
Specialized accounts are a higher protection level suitable for sensitive users. Because of their higher business
impact, specialized accounts warrant additional monitoring and prioritization during security alerts, incident
investigations, and threat hunting.
Specialized security builds on the strong MFA in enterprise security by identifying the most sensitive accounts
and ensuring alerts and response processes are prioritized:
1. Identify Sensitive Accounts - See specialized security level guidance for identifying these accounts.
2. Tag Specialized Accounts - Ensure each sensitive account is tagged
a. Configure Microsoft Sentinel Watchlists to identify these sensitive accounts
b. Configure Priority Account Protection in Microsoft Defender for Office 365 and designate specialized
and privileged accounts as priority accounts -
3. Update Security Operations processes - to ensure these alerts are given the highest priority
4. Set up Governance - Update or create governance process to ensure that
a. All new roles to are evaluated for specialized or privileged classifications as they are created or
changed
b. All new accounts are tagged as they are created
c. Continuous or periodic out of band checks to ensure that roles and accounts didn't get missed by
normal governance processes.
Privileged accounts
Privileged accounts have the highest level of protection because they represent a significant or material
potential impact on the organization's operations if compromised.
Privileged accounts always include IT Admins with access to most or all enterprise systems, including most or all
business critical systems. Other accounts with a high business impact may also warrant this additional level of
protection. For more information about which roles and accounts should be protected at what level, see the
article Privileged Security.
In addition to specialized security , privileged account security increases both:
Prevention - add controls to restrict the usage of these accounts to the designated devices, workstations, and
 intermediaries.
Response - closely monitor these accounts for anomalous activity and rapidly investigate and remediate the
risk.

Configuring privileged account security

Follow the guidance in the Security rapid modernization plan to both increase the security of your privileged
accounts and decrease your cost to manage.

Next steps
Securing privileged access overview
Privileged access strategy
Measuring success
Security levels
Intermediaries
Interfaces
Privileged access devices
Enterprise access model
 Privileged access: Intermediaries
12/13/2021 • 12 minutes to read • Edit Online

Security of intermediary devices is a critical component of securing privileged access.

Intermediaries add link to the chain of Zero Trust assurance for the user or administrator's end to end session, so
they must sustain (or improve) the Zero Trust security assurances in the session. Examples of intermediaries
include virtual private networks (VPNs), jump servers, virtual desktop infrastructure (VDI), as well as application
publishing through access proxies.

An attacker can attack an intermediary to attempt to escalating privileges using credentials stored on them, get
network remote access to corporate networks, or exploit trust in that device if being used for Zero Trust access
decisions. Targeting intermediaries has become an all too common, especially for organizations that don't
rigorously maintain the security posture of these devices. For example, credentials collected from VPN devices.

Intermediaries vary in purpose and technology, but typically provide remote access, session security, or both:
Remote access - Enable access to systems on enterprise networks from the internet
Session security - Increase security protections and visibility for a session
 Unmanaged device scenario - Providing a managed virtual desktop to be accessed by unmanaged
devices (for example, personal employee devices) and/or devices managed by a partner/vendor.
Administrator security scenario - Consolidate administrative pathways and/or increase security
with just in time access, session monitoring and recording, and similar capabilities.
Ensuring security assurances are sustained from the originating device and account through to the resource
interface requires understanding the risk profile of the intermediary and mitigation options.

Attacker opportunity and value

Different intermediary types perform unique functions so they each require a different security approach,
though there are some critical commonalities like rapidly applying security patches to appliances, firmware,
operating systems, and applications.

The attacker oppor tunity is represented by the available attack surface an attack operator can target:
Native cloud ser vices like Azure AD PIM, Azure Bastion, and Azure AD App Proxy offer a limited attack
surface to attackers. While they are exposed to the public internet, customers (and attackers) have no access
to underlying operating systems providing the services and they are typically maintained and monitored
consistently via automated mechanisms at the cloud provider. This smaller attack surface limits the available
options to attackers vs. classic on-premises applications and appliances that must be configured, patched,
and monitored by IT personnel who are often overwhelmed by conflicting priorities and more security tasks
than they have time to complete.
Vir tual Private Networks (VPNs) and Remote Desktops / Jump ser vers frequently have a significant
attacker opportunity as they are exposed to the internet to provide remote access and the maintenance of
these systems is frequently neglected. While they only have a few network ports exposed, attackers only
need access to one unpatched service for an attack.
Third-par ty PIM/PAM services are frequently hosted on-premises or as a VM on Infrastructure as a Service
(IaaS) and are typically only available to intranet hosts. While not directly internet exposed, a single
compromised credential may allow attackers to access the service over VPN or another remote access
medium.
Attacker value represents what an attacker can gain by compromising an intermediary. A compromise is
defined as an attacker gaining full control over the application/VM and/or an administrator of the customer
instance of the cloud service.
The ingredients that attackers can collect from an intermediary for the next stage of their attack include:
Get network connectivity to communicate with most or all resource on enterprise networks. This access is
 typically provided by VPNs and Remote Desktop / Jump server solutions. While Azure Bastion and Azure AD
App Proxy (or similar third-party solutions) solutions also provide remote access, these solutions are
typically application or server-specific connections and don’t provide general network access
Impersonate device identity - can defeat Zero Trust mechanisms if a device is required for authentication
and/or be used by an attacker to gather intelligence on the targets networks. Security Operations teams
often don't closely monitor device account activity and focus only on user accounts.
Steal account credentials to authenticate to resources, which are the most valuable asset to attackers as it
offers the ability to elevate privileges to access their ultimate goal or the next stage in the attack. Remote
Desktop / Jump servers and third-party PIM/PAM are the most attractive targets and have the “All your eggs
in one basket” dynamic with increased attacker value and security mitigations:
PIM/PAM solutions typically store the credentials for most or all privileged roles in the organization,
making them a highly lucrative target to compromise or to weaponize.
Azure AD PIM doesn't offer attackers the ability to steal credentials because it unlocks privileges
already assigned to an account using MFA or other workflows, but a poorly designed workflow could
allow an adversary to escalate privileges.
Remote Desktop / Jump ser vers used by administrators provide a host where many or all
sensitive sessions pass through, enabling attackers to use standard credential theft attack tools to steal
and reuse these credentials.
VPNs can store credentials in the solution, providing attackers with a potential treasure trove of
privilege escalation, leading to the strong recommendation to use Azure AD for authentication to
mitigate this risk.

Intermediary security profiles

Establishing these assurances requires a combination of security controls, some of which are common to many
intermediaries, and some of which specific to the type of intermediary.

An intermediary is a link in the Zero Trust chain that presents an interface to users/devices and then enables
access to the next interface. The security controls must address inbound connections, security of the
intermediary device/application/service itself, and (if applicable) provide Zero Trust security signals for the next
interface.

Common security controls

The common security elements for intermediaries are focused on maintaining good security hygiene for
enterprise and specialized levels, with additional restrictions for privilege security.
These security controls should be applied to all types of intermediaries:
Enforce inbound connection security - Use Azure AD and Conditional Access to ensure all inbound
connections from devices and accounts are known, trusted, and allowed. For more information, see the
article Secuiting privileged interfaces for detailed definitions for device and account requirements for
enterprise and specialized.
Proper system maintenance - All intermediaries must follow good security hygiene practices including:
Secure configuration - Follow manufacturer or industry security configuration baselines and best
practices for both the application and any underlying operating systems, cloud services, or other
dependencies. Applicable guidance from Microsoft includes the Azure Security Baseline and Windows
Baselines.
Rapid patching - Security updates and patches from the vendors must be applied rapidly after
release.
Role-Based Access Control (RBAC) models can be abused by attackers to escalate privileges. The RBAC
model of the intermediary must be carefully review to ensure that only authorized personnel that are
protected at a specialized or privileged level are granted administrative privileges. This model must include
any underlying operating systems or cloud services (root account password, local administrator
users/groups, tenant administrators, etc.).
Endpoint detection and response (EDR) and outbound trust signal - Devices that include a full
operating system should be monitored and protected with an EDR like Microsoft Defender for Endpoint. This
control should be configured to provides device compliance signals to Conditional Access so that policy can
enforce this requirement for interfaces.
Privileged Intermediaries require additional security controls:
Role-Based Access Control (RBAC) - Administrative rights must be restricted to only privileged roles
meeting that standard for workstations and accounts.
Dedicated devices (optional) - because of the extreme sensitivity of privileged sessions, organizations
may choose to implement dedicated instances of intermediary functions for privileged roles. This control
enables additional security restrictions for these privileged intermediaries and closer monitoring of
privileged role activity.

Security guidance for each intermediary type

This section contains specific security guidance unique to each type of intermediary.
Privileged Access Management / Privileged Identity management
One type of intermediary designed explicitly for security use cases is privileged identity management /
privileged access management (PIM/PAM) solutions.
Use cases and scenarios for PIM/PAM
PIM/PAM solutions are designed to increase security assurances for sensitive accounts that would be covered by
specialized or privileged profiles, and typically focus first on IT administrators.
While features vary between PIM/PAM vendors, many solutions provide security capabilities to:
 Simplify service account management and password rotation (a critically important capability)
Provide advanced workflows for just in time (JIT) access
Record and monitor administrative sessions

IMPORTANT
PIM/PAM capabilities provide excellent mitigations for some attacks, but do not address many privielged access
risks, notably risk of device compromise. While some vendors advocate that their PIM/PAM solution is a 'silver
bullet' solution that can mitigate device risk, our experience investigating customer incidents has consistently
proven that this does not work in practice.
An attacker with control of a workstation or device can use those credentials (and privileges assigned to them)
while the user is logged on (and can often steal credentials for later use as well). A PIM/PAM solution alone cannot
consistently and reliably see and mitigate these device risks, so you must have discrete device and account
protections that complement each other.

Security risks and recommendations for PIM/PAM

The capabilities from each PIM/PAM vendor vary on how to secure them, so review and follow your vendor's
specific security configuration recommendations and best practices.

NOTE
Ensure you set up a second person in business critical workflows to help mitigate insider risk (increases the cost/friction
for potential collusion by insider threats).

End-user Virtual Private Networks

Virtual Private Networks (VPNs) are intermediaries that provide full network access for remote endpoints,
typically require the end user to authenticate, and can store credentials locally to authenticate inbound user
sessions.

NOTE
This guidance refers only to "point to site" VPNs used by users, not "site to site" VPNs that are typically used for
datacenter/application connectivity.

Use cases and scenarios for VPNs

VPNs establish remote connectivity to enterprise network to enable resource access for users and
administrators.
Security risks and recommendations for VPNs
The most critical risks to VPN intermediaries are from maintenance neglect, configuration issues, and local
storage of credentials.
Microsoft recommends a combination of controls for VPN intermediaries:
Integrate Azure AD authentication - to reduce or eliminate risk of locally stored credentials (and any
overhead burden to maintain them) and enforce Zero Trust policies on inbound accounts/devices with
conditional access. For guidance on integrating, see
Azure VPN AAD integration
Enable Azure AD Authentication on the VPN gateway
Integrating third-party VPNs
Cisco AnyConnect
 Palo Alto Networks GlobalProtect and Captive Portal
F5
Fortinet FortiGate SSL VPN
Citrix NetScaler
Zscaler Private Access (ZPA)
and more
Rapid patching - Ensure that all organizational elements support rapid patching including:
Organizational sponsorship and leadership support for requirement
Standard technical processes for updating VPNs with minimal or zero downtime. This process
should include VPN software, appliances, and any underlying operating systems or firmware
Emergency processes to rapidly deploy critical security updates
Governance to continually discover and remediate any missed items
Secure configuration - The capabilities from each VPN vendor vary on how to secure them, so review and
follow your vendor's specific security configuration recommendations and best practices
Go beyond VPN - Replace VPNs over time with more secure options like Azure AD App Proxy or Azure
Bastion as these provide only direct application/server access rather than full network access. Additionally
Azure AD App Proxy allows session monitoring for additional security with Microsoft Defender for Cloud
Apps.

Azure AD App Proxy

Azure AD App Proxy and similar third-party capabilities provide remote access to legacy and other applications
hosted on-premises or on IaaS VMs in the cloud.
Use cases and scenarios for Azure AD App Proxy
This solution is suitable for publishing legacy end-user productivity applications to authorized users over the
internet. It can also be used for publishing some administrative applications.
Security risks and recommendations for Azure AD App Proxy
Azure AD App proxy effectively retrofits modern Zero Trust policy enforcement to existing applications. For more
information, see Security considerations for Azure AD Application Proxy
Azure AD Application Proxy can also integrate with Microsoft Defender for Cloud Apps to add Conditional
Access App Control session security to:
Prevent data exfiltration
Protect on download
Prevent upload of unlabeled files
 Monitor user sessions for compliance
Block access
Block custom activities
For more information, see Deploy Defender for Cloud Apps Conditional Access App Control for Azure AD apps
As you publish applications via the Azure AD Application Proxy, Microsoft recommends having application
owners work with security teams to follow least privilege and ensure access to each application is made
available to only the users that require it. As you deploy more apps this way, you may be able to offset some
end-user point to site VPN usage.
Remote Desktop / jump server
This scenario provides a full desktop environment running one or more applications. This solution has a number
of different variations including:
Experiences - Full desktop in a window or a single application projected experience
Remote host - may be a shared VM or a dedicated desktop VM using Windows Virtual Desktop (WVD) or
another Virtual Desktop Infrastructure (VDI) solution.
Local device - may be a mobile device, a managed workstation, or a personal/partner managed
workstation
Scenario - focused on user productivity applications or on administrative scenarios, often called a 'jump
server'
Use cases and security recommendations for Remote Desktop / Jump server
The most common configurations are:
Direct Remote Desktop Protocol (RDP) - This configuration is not recommended for internet connections
because RDP is a protocol that has limited protections against modern attacks like password spray. Direct
RDP should be converted to either:
RDP through a gateway published by Azure AD App Proxy
Azure Bastion
RDP through a gateway using
Remote Desktop Services (RDS) included in Windows Server. Publish with Azure AD Application Proxy.
Windows Virtual Desktop (WVD) - Follow Windows Virtual Desktop security best practices.
Third-party VDI - Follow manufacturer or industry best practices, or adapt WVD guidance to your
solution
Secure Shell (SSH) server - providing remote shell and scripting for technology departments and workload
owners. Securing this configuration should include:
Following industry/manufacturer best practices to securely configure it, change any default passwords
(if applicable), and using SSH keys instead of passwords, and securely storing and managing SSH
keys.
Use Azure Bastion for SSH remoting to resources hosted in Azure - Connect to a Linux VM using
Azure Bastion
Azure Bastion
Azure Bastion is an intermediary that is designed to provide secure access to Azure resources using a browser
and the Azure portal. Azure Bastion provides access resources in Azure that support Remote Desktop Protocol
(RDP) and Secure Shell (SSH) protocols.
Use cases and scenarios for Azure Bastion
Azure Bastion effectively provides a flexible solution that can be used by IT Operations personnel and workload
administrators outside of IT to manage resources hosted in Azure without requiring a full VPN connection to the
environment.
Security risks and recommendations for Azure Bastion
Azure Bastion is accessed through the Azure portal, so ensure that your Azure portal interface requires the
appropriate level of security for the resources in it and roles using it, typically privileged or specialized level.
Additional guidance is available in the Azure Bastion Documentation

Next steps
Securing privileged access overview
Privileged access strategy
Measuring success
Security levels
Privileged access accounts
Interfaces
Privileged access devices
Enterprise access model
 Privileged access: Interfaces
12/13/2021 • 4 minutes to read • Edit Online

A critical component of securing privileged access is the application of zero trust policy to ensure that devices,
accounts, and intermediaries meet security requirements before providing access.
This policy ensures users and devices initiating the inbound session are known, trusted, and allowed to access
the resource (via the interface). The policy enforcement is performed by the Azure AD Conditional Access policy
engine that evaluates policy assigned to the specific application interface (such as Azure portal, Salesforce,
Office 365, AWS, Workday, and others).

This guidance defines three security levels for interface security that you can use for assets with different
sensitivity levels. These levels are configured in the securing privileged access rapid modernization plan (RAMP)
and correspond to security levels of accounts and devices.
The security requirements for inbound sessions to interfaces apply to accounts and the source device, whether
it’s a direct connection from physical devices or a Remote Desktop / Jump server intermediary. Intermediaries
can accept sessions from personal devices to provide enterprise security level (for some scenarios), but
specialized or privileged intermediaries should not allow connections from lower levels because of the security
sensitive nature of their roles.

NOTE
These technologies provide strong end to end access control to the application interface, but the resource itself must also
be secured from out of band attacks on the application code/functionality, unpatched vulnerabilities or configuration
errors in the underlying operating system or firmware, on data at rest or in transit, supply chains, or other means.
Ensure to assess and discover risks to the assets themselves for complete protection. Microsoft provides tooling and
guidance to help you with that including Microsoft Defender for Cloud, Microsoft Secure Score, and threat modelling
guidance.

Interface examples
Interfaces come in different forms, typically as:
 Cloud service/application websites such as Azure portal, AWS, Office 365
Desktop Console managing an on-premises application (Microsoft Management Console (MMC) or custom
application)
Scripting/Console Interface such as Secure Shell (SSH) or PowerShell
While some of these directly support Zero Trust enforcement via the Azure AD Conditional Access policy engine,
some of them will need to be published via an intermediary such as Azure AD App Proxy or Remote Desktop /
jump server.

Interface security
The ultimate goal of interface security is to ensure that each inbound session to the interface is known, trusted,
and allowed:
Known – User is authenticated with strong authentication and device is authenticated (with exceptions for
personal devices using a Remote Desktop or VDI solution for enterprise access)
Trusted – Security health is explicitly validated and enforced for accounts and devices using a Zero Trust
policy engine
Allowed – Access to the resources follows least privilege principle using a combination of controls to ensure
it can only be accessed
By the right users
At the right time ( just in time access, not permanent access)
With the right approval workflow (as needed)
At an acceptable risk/trust level

Interface security controls

Establishing interface security assurances requires a combination of security controls including:
Zero Trust policy enforcement - using Conditional Access to ensure that the inbound sessions meet the
requirements for:
Device Trust to ensure the device at minimum:
Is managed by the enterprise
Has endpoint detection and response on it
Is compliant with organizations configuration requirements
Isn't infected or under attack during the session
User Trust is high enough based on signals including:
Multi-factor authentication usage during initial logon (or added later to increase trust)
Whether this session matches historical behavior patterns
Whether the account or current session triggers any alerts based on threat intelligence
Azure AD Identity Protection risk
Role-based access control (RBAC) model that combines enterprise directory groups/permissions and
application-specific roles, groups, and permissions
Just in time access workflows that ensure specific requirements for privileges (peer approvals, audit trail,
privileged expiration, etc.) are enforced before allowing privileges the account is eligible for.

Interface security levels

This guidance defines three levels of security. For more information on these levels, see Keep it Simple -
Personas and Profiles. For implementation guidance, see the rapid modernization plan.
Enterprise interface
Enterprise interface security is suitable for all enterprise users and productivity scenarios. Enterprise also serves
as a starting point for higher sensitivity workloads that you can incrementally build on to reach specialized and
privileged access levels of assurance.
Zero Trust policy enforcement - on inbound sessions using Conditional Access to ensure that users and
devices are secured at the enterprise or higher level
To support, bring your own device (BYOD) scenarios, personal devices, and partner-managed devices
may be allowed connect if they use an enterprise intermediary such as a dedicated Windows Virtual
Desktop (WVD) or similar Remote Desktop / Jump server solution.
Role-Based Access Control (RBAC) - Model should ensure that the application is administered only by roles at
the specialized or privileged security level
Specialized interface
Security controls for specialized interfaces should include
Zero Trust policy enforcement - on inbound sessions using Conditional Access to ensure that users and
devices are secured at the specialized or privileged level
Role-Based Access Control (RBAC) - Model should ensure that the application is administered only by roles at
the specialized or privileged security level
Just in time access workflows (optional) - that enforce least privilege by ensuring privileges are used only by
authorized users during the time they are needed.
Privileged interface
Security controls for specialized interfaces should include
Zero Trust policy enforcement - on inbound sessions using Conditional Access to ensure that users and
devices are secured at the privileged level
Role-Based Access Control (RBAC) - Model should ensure that the application is administered only by roles at
the privileged security level
Just in time access workflows (required) that enforce least privilege by ensuring privileges are used only by
authorized users during the time they are needed.

Next steps
Securing privileged access overview
Privileged access strategy
Measuring success
Security levels
Privileged access accounts
Intermediaries
Privileged access devices
Enterprise access model
 Securing devices as part of the privileged access
story
12/13/2021 • 6 minutes to read • Edit Online

This guidance is part of a complete privileged access strategy and is implemented as part of the Privileged
access deployment
End to end zero trust security for privileged access requires a strong foundation of device security upon which
to build other security assurances for the session. While security assurances may be enhanced in the session,
they will always be limited by how strong the security assurances are in the originating device. An attacker with
control of this device can impersonate users on it or steal their credentials for future impersonation. This risk
undermines other assurances on the account, intermediaries like jump servers, and on the resources
themselves. For more information, see clean source principle
The article provides an overview of security controls to provide a secure workstation for sensitive users
throughout its lifecycle.

This solution relies on core security capabilities in the Windows 10 operating system, Microsoft Defender for
Endpoint, Azure Active Directory, and Microsoft InTune.

Who benefits from a secure workstation?

All users and operators benefit from using a secure workstation. An attacker who compromises a PC or device
can impersonate or steal credentials/tokens for all accounts that use it, undermining many or all other security
assurances. For administrators or sensitive accounts, this allows attackers to escalate privileges and increase the
access they have in your organization, often dramatically to domain, global, or enterprise administrator
privileges.
For details on security levels and which users should be assigned to which level, see Privileged access security
levels

Device Security Controls

The successful deployment of a secure workstation requires it to be part of an end to end approach including
devices, accounts, intermediaries, and security policies applied to your application interfaces. All elements of the
stack must be addressed for a complete privileged access security strategy.
This table summarizes the security controls for different device levels:

P RO F IL E EN T ERP RISE SP EC IA L IZ ED P RIVIL EGED

Microsoft Endpoint Yes Yes Yes

Manager (MEM) managed

Deny BYOD Device No Yes Yes

enrollment

MEM security baseline Yes Yes Yes

applied

Microsoft Defender for Yes* Yes Yes

Endpoint

Join personal device via Yes* Yes* No

Autopilot

URLs restricted to approved Allow Most Allow Most Deny Default

list

Removal of admin rights Yes Yes

Application execution Audit -> Enforced Yes

control (AppLocker)

Applications installed only Yes Yes

by MEM

NOTE
The solution can be deployed with new hardware, existing hardware, and bring your own device (BYOD) scenarios.

At all levels, good security maintenance hygiene for security updates will be enforced by Intune policies. The
differences in security as the device security level increases are focused on reducing the attack surface that an
attacker can attempt to exploit (while preserving as much user productivity as possible). Enterprise and
specialized level devices allow productivity applications and general web browsing, but privileged access
workstations do not. Enterprise users may install their own applications, but specialized users may not (and are
not local administrators of their workstations).

NOTE
Web browsing here refers to general access to arbitrary websites which can be a high risk activity. Such browsing is
distinctly different from using a web browser to access a small number of well-known administrative websites for services
like Azure, Microsoft 365, other cloud providers, and SaaS applications.

Hardware root of trust

Essential to a secured workstation is a supply chain solution where you use a trusted workstation called the 'root
of trust'. Technology that must be considered in the selection of the root of trust hardware should include the
following technologies included in modern laptops:
 Trusted Platform Module (TPM) 2.0
BitLocker Drive Encryption
UEFI Secure Boot
Drivers and Firmware Distributed through Windows Update
Virtualization and HVCI Enabled
Drivers and Apps HVCI-Ready
Windows Hello
DMA I/O Protection
System Guard
Modern Standby
For this solution, root of trust will be deployed using Windows Autopilot technology with hardware that meets
the modern technical requirements. To secure a workstation, Autopilot lets you leverage Microsoft OEM-
optimized Windows 10 devices. These devices come in a known good state from the manufacturer. Instead of
reimaging a potentially insecure device, Autopilot can transform a Windows 10 device into a “business-ready”
state. It applies settings and policies, installs apps, and even changes the edition of Windows 10.

Device roles and profiles

This guidance shows how to harden Windows 10 and reduce the risks associated with device or user
compromise. To take advantage of the modern hardware technology and root of trust device, the solution uses
Device Health Attestation. This capability is present to ensure the attackers cannot be persistent during the early
boot of a device. It does so by using policy and technology to help manage security features and risks.
 Enterprise Device – The first managed role is good for home users, small business users, general
developers, and enterprises where organizations want to raise the minimum security bar. This profile permits
users to run any applications and browse any website, but an anti-malware and endpoint detection and
response (EDR) solution like Microsoft Defender for Endpoint is required. A policy-based approach to
increase the security posture is taken. It provides a secure means to work with customer data while also
using productivity tools like email and web browsing. Audit policies and Intune allow you to monitor an
Enterprise workstation for user behavior and profile usage.
The enterprise security profile in the privileged access deployment guidance uses JSON files to configure this
with Windows 10 and the provided JSON files.
Specialized Device – This represents a significant step up from enterprise usage by removing the ability to
self-administer the workstation and limiting which applications may run to only the applications installed by
an authorized administrator (in the program files and pre-approved applications in the user profile location.
Removing the ability to install applications may impact productivity if implemented incorrectly, so ensure
that you have provided access to Microsoft store applications or corporate managed applications that can be
rapidly installed to meet users needs. For guidance on which users should be configured with specialized
level devices, see Privileged access security levels
The Specialized security user demands a more controlled environment while still being able to do
activities such as email and web browsing in a simple-to-use experience. These users expect features
such as cookies, favorites, and other shortcuts to work but do not require the ability to modify or
debug their device operating system, install drivers, or similar.
The specialized security profile in the privileged access deployment guidance uses JSON files to configure this
with Windows 10 and the provided JSON files.
Privileged Access Workstation (PAW) – This is the highest security configuration designed for extremely
sensitive roles that would have a signficant or material impact on the organization if their account was
compromised. The PAW configuration includes security controls and policies that restrict local administrative
access and productivity tools to minimize the attack surface to only what is absolutely required for
performing sensitive job tasks. This makes the PAW device difficult for attackers to compromise because it
blocks the most common vector for phishing attacks: email and web browsing. To provide productivity to
these users, separate accounts and workstations must be provided for productivity applications and web
browsing. While inconvenient, this is a necessary control to protect users whose account could inflict damage
to most or all resources in the organization.
A Privileged workstation provides a hardened workstation that has clear application control and
application guard. The workstation uses credential guard, device guard, app guard, and exploit guard
 to protect the host from malicious behavior. All local disks are encrypted with BitLocker and web
traffic is restricted to a limit set of permitted destinations (Deny all).
The privileged security profile in the privileged access deployment guidance uses JSON files to configure this
with Windows 10 and the provided JSON files.

Next steps
Deploy a secure Azure-managed workstation.
 Enterprise access model
12/13/2021 • 3 minutes to read • Edit Online

This document describes an overall enterprise access model that includes context of how a privileged access
strategy fits in. For a roadmap on how to adopt a privileged access strategy, see the rapid modernization plan
(RaMP). For implementation guidance to deploy this, see privileged access deployment
Privileged access strategy is part of an overall enterprise access control strategy. This enterprise access model
shows how privileged access fits into an overall enterprise access model.
The primary stores of business value that an organization must protect are in the Data/Workload plane:

The applications and data typically store a large percentage of an organization's:

Business processes in applications and workloads
Intellectual proper ty in data and applications
The enterprise IT organization manages and supports the workloads and the infrastructure they are hosted on,
whether it's on-premises, on Azure, or a third-party cloud provider, creating a management plane . Providing
consistent access control to these systems across the enterprise requires a control plane based on centralized
enterprise identity system(s), often supplemented by network access control for older systems like operational
technology (OT) devices.

Each of these planes has control of the data and workloads by virtue of their functions, creating an attractive
pathway for attackers to abuse if they can gain control of either plane.
For these systems to create business value, they must be accessible to internal users, partners, and customers
using their workstations or devices (often using remote access solutions) - creating user access pathways. They
must also frequently be available programmatically via application programming interfaces (APIs) to facilitate
process automation, creating application access pathways.
Finally, these systems must be managed and maintained by IT staff, developers, or others in the organizations,
creating privileged access pathways. Because of the high level of control they provide over business critical
assets in the organization, these pathways must be stringently protected against compromise.

Providing consistent access control in the organization that enables productivity and mitigates risk requires you
to
Enforce Zero Trust principles on all access
Assume Breach of other components
Explicit validation of trust
Least privilege access
Pervasive security and policy enforcement across
Internal and external access to ensure consistent policy application
 All access methods including users, admins, APIs, service accounts, etc.
Mitigate unauthorized privilege escalation
Enforce hierarchy – to prevent control of higher planes from lower planes (via attacks or abuse of
legitimate processes)
Control plane
Management plane
Data/workload plane
Continuously audit for configuration vulnerabilities enabling inadvertent escalation
Monitor and respond to anomalies that could represent potential attacks

Evolution from the legacy AD tier model

The enterprise access model supersedes and replaces the legacy tier model that was focused on containing
unauthorized escalation of privilege in an on-premises Windows Server Active Directory environment.

The enterprise access model incorporates these elements as well as full access management requirements of a
modern enterprise that spans on-premises, multiple clouds, internal or external user access, and more.

Tier 0 scope expansion

Tier 0 expands to become the control plane and addresses all aspects of access control, including networking
where it is the only/best access control option, such as legacy OT options
Tier 1 splits
To increase clarity and actionability, what was tier 1 is now split into the following areas:
 Management plane – for enterprise-wide IT management functions
Data/Workload plane – for per-workload management, which is sometimes performed by IT personnel
and sometimes by business units
This split ensures focus for protecting business critical systems and administrative roles that have high intrinsic
business value, but limited technical control. Additionally, this split better accommodates developers and
DevOps models vs. focusing too heavily on classic infrastructure roles.
Tier 2 splits
To ensure coverage for application access and the various partner and customer models, Tier 2 was split into the
following areas:
User access – which includes all B2B, B2C, and public access scenarios
App access – to accommodate API access pathways and resulting attack surface

Next steps
Securing privileged access overview
Privileged access strategy
Measuring success
Security levels
Privileged access accounts
Intermediaries
Interfaces
Privileged access devices
 Privileged access deployment
12/13/2021 • 23 minutes to read • Edit Online

This document will guide you through implementing the technical components of the privileged access strategy,
including secure accounts, workstations and devices, and interface security (with conditional access policy).

This guidance sets up all of the profiles for all three security levels and should be assigned your organizations
roles based on the Privileged access security levels guidance. Microsoft recommends configuring them in the
order described in the rapid modernization plan (RAMP)

License requirements
The concepts covered in this guide assume you have Microsoft 365 Enterprise E5 or an equivalent SKU. Some of
the recommendations in this guide can be implemented with lower SKUs. For more information, see Microsoft
365 Enterprise licensing.
To automate license provisioning, consider group-based licensing for your users.

Azure Active Directory configuration

Azure Active Directory (Azure AD) manages users, groups, and devices for your administrator workstations.
Enable identity services and features with an administrator account.
When you create the secured workstation administrator account, you expose the account to your current
workstation. Make sure you use a known safe device to do this initial configuration and all global configuration.
To reduce the attack exposure for the first-time experience, consider following the guidance to prevent malware
infections.
Require multi-factor authentication, at least for your administrators. See Conditional Access: Require MFA for
administrators for implementation guidance.
Azure AD users and groups
1. From the Azure portal, browse to Azure Active Director y > Users > New user .
2. Create your device user by following the steps in the create user tutorial.
3. Enter:
Name - Secure Workstation Administrator
User name - [email protected]
Director y role - Limited administrator and select the Intune Administrator role.
 Usage Location - United Kingdom
4. Select Create .
Create your device administrator user.
1. Enter:
Name - Secure Workstation Administrator
User name - [email protected]
Director y role - Limited administrator and select the Intune Administrator role.
2. Select Create .
Next, you create four groups: Secure Workstation Users , Secure Workstation Admins , Emergency
BreakGlass and Secure Workstation Devices .
From the Azure portal, browse to Azure Active Director y > Groups > New group .
1. For the workstation users group, you might want to configure group-based licensing to automate
provisioning of licenses to users.
2. For the workstation users group, enter:
Group type - Security
Group name - Secure Workstation Users
Membership type - Assigned
3. Add your secure workstation user: [email protected]
4. You can add any other users that will be using secure workstations.
5. Select Create .
6. For the Privileged Workstation Admins group, enter:
Group type - Security
Group name - Secure Workstation Admins
Membership type - Assigned
7. Add your secure workstation user: [email protected]
8. You can add any other users that will be managing secure workstations.
9. Select Create .
10. For the Emergency BreakGlass group, enter:
Group type - Security
Group name - Emergency BreakGlass
Membership type - Assigned
11. Select Create .
12. Add Emergency Access accounts to this group.
13. For the workstation devices group, enter:
Group type - Security
Group name - Secure Workstations
Membership type - Dynamic Device
Dynamic Membership rules - (device.devicePhysicalIds -any _ -contains "[OrderID]:PAW")
14. Select Create .
Azure AD device configuration
Specify who can join devices to Azure AD
Configure your devices setting in Active Directory to allow your administrative security group to join devices to
your domain. To configure this setting from the Azure portal:
1. Go to Azure Active Director y > Devices > Device settings .
2. Choose Selected under Users may join devices to Azure AD , and then select the "Secure Workstation
Users" group.
Remove local admin rights
This method requires that users of the VIP, DevOps, and Privileged workstations have no administrator rights on
their machines. To configure this setting from the Azure portal:
1. Go to Azure Active Director y > Devices > Device settings .
2. Select None under Additional local administrators on Azure AD joined devices .
Refer to How to manage the local administrators group on Azure AD joined devices for details on how to
manage members of the local administrators group.
Require multi-factor authentication to join devices
To further strengthen the process of joining devices to Azure AD:
1. Go to Azure Active Director y > Devices > Device settings .
2. Select Yes under Require Multi-Factor Auth to join devices .
3. Select Save .
Configure mobile device management
From the Azure portal:
1. Browse to Azure Active Director y > Mobility (MDM and MAM) > Microsoft Intune .
2. Change the MDM user scope setting to All .
3. Select Save .
These steps allow you to manage any device with Microsoft Endpoint Manager. For more information, see Intune
Quickstart: Set up automatic enrollment for Windows 10 devices. You create Intune configuration and
compliance policies in a future step.
Azure AD Conditional Access
Azure AD Conditional Access can help restrict privileged administrative tasks to compliant devices. Predefined
members of the Secure Workstation Users group are required to perform multi-factor authentication when
signing in to cloud applications. A best practice is to exclude emergency access accounts from the policy. For
more information, see Manage emergency access accounts in Azure AD.
Conditional Access only allowing secured workstation ability to access Azure portal
Organizations should block Privileged Users from being able to connect to cloud management interfaces,
portals and PowerShell, from non-PAW devices.
To block unauthorized devices from being able to access cloud management interfaces, follow the guidance in
the article Conditional Access: Filters for Devices (preview). It's essential that while deploying this feature you
consider, emergency access account functionality. These accounts should be used only for extreme cases and the
account managed through policy.

NOTE
You will need to create a user group, and include your emergency user that can bypass the Conditional Access policy. For
our example we have a security group called Emergency BreakGlass
This policy set will ensure that your Administrators must use a device that is able to present a specific device
attribute value, that that MFA is satisfied, and the device is marked as compliant by Microsoft Endpoint Manager
and Microsoft Defender for Endpoint.
Organizations should also consider blocking legacy authentication protocols in their environments. There are
multiple ways to accomplish this task, for more information about blocking legacy authentication protocols, see
the article, How to: Block legacy authentication to Azure AD with Conditional Access.

Microsoft Intune configuration

Device enrollment deny BYOD
In our sample, we recommend that BYOD devices not be permitted. Using Intune BYOD enrollment allows users
to enroll devices that are less, or not trusted. However it's important to note that in organizations that have a
limited budget to purchase new devices, looking to use existing hardware fleet, or considering non-windows
devices, might consider the BYOD capability in Intune to deploy the Enterprise profile.
The following guidance will configure Enrollment for deployments that will deny BYOD access.
Set enrollment restrictions preventing BYOD
1. In the Microsoft Endpoint Manager admin center, choose > Devices > Enrollment restrictions > choose
the default restriction All Users
2. Select Proper ties > Platform settings Edit
3. Select Block for All types, except Windows MDM.
4. Select Block for all Personally owned items.
Create an Autopilot deployment profile
After creating a device group, you must create a deployment profile to configure the Autopilot devices.
1. In the Microsoft Endpoint Manager admin center, choose Device enrollment > Windows enrollment
> Deployment Profiles > Create Profile .
2. Enter:
Name - Secure workstation deployment profile .
Description - Deployment of secure workstations .
Set Conver t all targeted devices to Autopilot to Yes . This setting makes sure that all devices in
the list get registered with the Autopilot deployment service. Allow 48 hours for the registration to be
processed.
3. Select Next .
For Deployment mode , choose Self-Deploying (Preview) . Devices with this profile are associated
with the user who enrolls the device. During the deployment, it is advisable to use the Self-
Deployment mode features to include:
Enrolls the device in Intune Azure AD automatic MDM enrollment, and only allow for a device
to be accessed until all policies, applications, certificates, and networking profiles are
provisioned on the device.
User credentials are required to enroll the device. It's essential to note that deploying a device
in the Self-Deploying mode will allow you to deploy laptops in a shared model. No user
assignment will happen until the device is assigned to a user for the first time. As a result, any
user policies such as BitLocker will not be enabled until a user assignment is completed. For
more information about how to log on to a secured device, see selected profiles.
Select your Language (Region), User account type standard .
4. Select Next .
 Select a scope tag if you have preconfigured one.
5. Select Next .
6. Choose Assignments > Assign to > Selected Groups . In Select groups to include , choose Secure
Workstations .
7. Select Next .
8. Select Create to create the profile. The Autopilot deployment profile is now available to assign to devices.
Device enrollment in Autopilot provides a different user experience based on device type and role. In our
deployment example, we illustrate a model where the secured devices are bulk deployed and can be shared, but
when used for the first time, the device is assigned to a user. For more information, see Intune Autopilot device
enrollment.
Enrollment Status Page
The Enrollment Status Page (ESP) displays provisioning progress after a new device is enrolled. To ensure that
devices are fully configured before use, Intune provides a means to Block device use until all apps and
profiles are installed .
Create and assign enrollment status page profile
1. In the Microsoft Endpoint Manager admin center, choose Devices > Windows > Windows enrollment >
Enrollment Status Page > Create profile .
2. Provide a Name and Description .
3. Choose Create .
4. Choose the new profile in the Enrollment Status Page list.
5. Set Show app profile installation progress to Yes .
6. Set Block device use until all apps and profiles are installed to Yes .
7. Choose Assignments > Select groups > choose Secure Workstation group > Select > Save .
8. Choose Settings > choose the settings you want to apply to this profile > Save .
Configure Windows Update
Keeping Windows 10 up to date is one of the most important things you can do. To maintain Windows in a
secure state, you deploy an update ring to manage the pace that updates are applied to workstations.
This guidance recommends that you create a new update ring and change the following default settings:
1. In the Microsoft Endpoint Manager admin center, choose Devices > Software updates > Windows 10
Update Rings .
2. Enter:
Name - Azure-managed workstation updates
Servicing channel - Semi-annual channel
Quality update deferral (days) - 3
Feature update deferral period (days) - 3
Automatic update behavior - Auto install and reboot without end-user control
Block user from pausing Windows updates - Block
Require user's approval to restart outside of work hours - Required
Allow user to restart (engaged restart) - Required
Transition users to engaged restart after an auto-restart (days) - 3
Snooze engaged restart reminder (days) - 3
Set deadline for pending restarts (days) - 3
3. Select Create .
4. On the Assignments tab, add the Secure Workstations group.
For more information about Windows Update policies, see Policy CSP - Update.
Microsoft Defender for Endpoint Intune integration
Microsoft Defender for Endpoint and Microsoft Intune work together to help prevent security breaches. They can
also limit the impact of breaches. ATP capabilities provide real-time threat detection as well as enable extensive
auditing and logging of the end-point devices.
To configure integration of Windows Defender for Endpoint and Microsoft Endpoint Manager:
1. In the Microsoft Endpoint Manager admin center, choose Endpoint Security > Microsoft Defender
ATP .
2. In step 1 under Configuring Windows Defender ATP , select Connect Windows Defender ATP to
Microsoft Intune in the Windows Defender Security Center .
3. In the Windows Defender Security Center:
a. Select Settings > Advanced features .
b. For Microsoft Intune connection , choose On .
c. Select Save preferences .
4. After a connection is established, return to Microsoft Endpoint Manager and select Refresh at the top.
5. Set Connect Windows devices version(20H2) 19042.450 and above to Windows Defender
ATP to On .
6. Select Save .
Create the device configuration profile to onboard Windows devices
1. Sign in to the Microsoft Endpoint Manager admin center, choose Endpoint security > Endpoint
detection and response > Create profile .
2. For Platform , select Windows 10 and Later .
3. For Profile type , select Endpoint detection and response , and then select Create .
4. On the Basics page, enter a PAW - Defender for Endpoint in the Name field and Description (optional) for
the profile, then choose Next .
5. On the Configuration settings page, configure the following option in Endpoint Detection and
Response :
Sample sharing for all files : Returns or sets the Microsoft Defender Advanced Threat Protection
Sample Sharing configuration parameter.
Onboard Windows 10 machines using Microsoft Endpoint Configuration Manager has more
details on these Microsoft Defender ATP settings.
6. Select Next to open the Scope tags page. Scope tags are optional. Select Next to continue.
7. On the Assignments page, select Secure Workstation group. For more information on assigning
profiles, see Assign user and device profiles.
Select Next .
8. On the Review + create page, when you're done, choose Create . The new profile is displayed in the list
when you select the policy type for the profile you created. OK , and then Create to save your changes,
which creates the profile.
For more information, see Windows Defender Advanced Threat Protection.
Finish workstation profile hardening
To successfully complete the hardening of the solution, download and execute the appropriate script. Find the
download links for your desired profile level :

P RO F IL E DO W N LO A D LO C AT IO N F IL EN A M E

Enterprise https://aka.ms/securedworkstationgit|
Enterprise-Workstation-Windows10-
(20H2).ps1

Specialized https://aka.ms/securedworkstationgit Specialized - Windows10-(20H2).ps1

Privileged https://aka.ms/securedworkstationgit Privileged-Windows10-(20H2).ps1

NOTE
The removal of of admin rights and access, as well as, Application execution control (AppLocker) are managed by the
policy profiles that are deployed.

After the script successfully executes, you can make updates to profiles and policies in Intune. The scripts will
create policies and profiles for you, but you must assign the policies to your Secure Workstations device
group.
Here's where you can find the Intune device configuration profiles created by the scripts: Azure por tal >
Microsoft Intune > Device configuration > Profiles .
Here's where you can find the Intune device compliance policies created by the scripts: Azure por tal >
Microsoft Intune > Device Compliance > Policies .
Run the Intune data export script DeviceConfiguration_Export.ps1 from the DeviceConfiguration GitHub
repository to export all current Intune profiles for comparison, and evaluation of the profiles.

Set rules in the Endpoint Protection Configuration Profile for

Microsoft Defender Firewall
Windows Defender Firewall policy settings are included in the Endpoint Protection Configuration Profile. The
behavior of the policy applied in described in the table below.

P RO F IL E IN B O UN D RUL ES O UT B O UN D RUL ES M ERGE B EH AVIO R

Enterprise Block Allow Allow

Specialized Block Allow Block

Privileged Block Block Block

Enterprise : This configuration is the most permissive as it mirrors the default behavior of a Windows Install. All
inbound traffic is blocked except for rules that are explicitly defined in the local policy rules as merging of local
rules is set to allowed. All outbound traffic is allowed.
Specialized : This configuration is more restrictive as it ignores all locally defined rules on the device. All
inbound traffic is blocked including locally defined rules the policy includes two rules to allow Delivery
Optimization to function as designed. All outbound traffic is allowed.
Privileged : All inbound traffic is blocked including locally defined rules the policy includes two rules to allow
Delivery Optimization to function as designed. Outbound traffic is also blocked apart from explicit rules that
allow DNS, DHCP, NTP, NSCI, HTTP, and HTTPS traffic. This configuration not only reduces the attack surface
presented by the device to the network it limits the outbound connections that the device can establish to only
those connections required to administer cloud services.

A P P L IC AT IO N REM OT E
RUL E DIREC T IO N A C T IO N / SERVIC E P ROTO C O L LO C A L P O RT S P O RT S

World Wide Outbound Allow All TCP All ports 80

Web Services
(HTTP Traffic-
out)

World Wide Outbound Allow All TCP All ports 443

Web Services
(HTTPS
Traffic-out)

Core Outbound Allow %SystemRoot TCP 546 547

Networking - %\system32\s
Dynamic Host vchost.exe
Configuration
Protocol for
IPv6(DHCPV6
-Out)

Core Outbound Allow Dhcp TCP 546 547

Networking -
Dynamic Host
Configuration
Protocol for
IPv6(DHCPV6
-Out)

Core Outbound Allow %SystemRoot TCP 68 67

Networking - %\system32\s
Dynamic Host vchost.exe
Configuration
Protocol for
IPv6(DHCP-
Out)

Core Outbound Allow Dhcp TCP 68 67

Networking -
Dynamic Host
Configuration
Protocol for
IPv6(DHCP-
Out)

Core Outbound Allow %SystemRoot UDP All Ports 53

Networking - %\system32\s
DNS (UDP- vchost.exe
Out)
 A P P L IC AT IO N REM OT E
RUL E DIREC T IO N A C T IO N / SERVIC E P ROTO C O L LO C A L P O RT S P O RT S

Core Outbound Allow Dnscache UDP All Ports 53

Networking -
DNS (UDP-
Out)

Core Outbound Allow %SystemRoot TCP All Ports 53

Networking - %\system32\s
DNS (TCP- vchost.exe
Out)

Core Outbound Allow Dnscache TCP All Ports 53

Networking -
DNS (TCP-
Out)

NSCI Probe Outbound Allow %SystemRoot TCP All ports 80

(TCP-Out) %\system32\s
vchost.exe

NSCI Probe - Outbound Allow NlaSvc TCP All ports 80

DNS (TCP-
Out)

Windows Outbound Allow %SystemRoot TCP All ports 80

Time (UDP- %\system32\s
Out) vchost.exe

Windows Outbound Allow W32Time UDP All ports 123

Time Probe -
DNS (UDP-
Out)

Delivery Inbound Allow %SystemRoot TCP 7680 All ports

Optimization %\system32\s
(TCP-In) vchost.exe

Delivery Inbound Allow DoSvc TCP 7680 All ports

Optimization
(TCP-In)

Delivery Inbound Allow %SystemRoot UDP 7680 All ports

Optimization %\system32\s
(UDP-In) vchost.exe

Delivery Inbound Allow DoSvc UDP 7680 All ports

Optimization
(UDP-In)

NOTE
There are two rules defined for each rule in the Microsoft Defender Firewall configuration. To restrict the inbound and
outbound rules to Windows Services, e.g. DNS Client, both the service name, DNSCache, and the executable path,
C:\Windows\System32\svchost.exe, need to be defined as separate rule rather than a single rule that is possible using
Group Policy.
You can make additional changes to the management of both inbound and outbound rules as needed for your
permitted and blocked services. For more information, see Firewall configuration service.
URL lock proxy
Restrictive URL traffic management includes:
Deny All outbound traffic except selected Azure and Microsoft services including Azure Cloud Shell and the
ability to allows self-service password reset.
The Privileged profile restricts the endpoints on the internet that the device can connect to using the
following URL Lock Proxy configuration.

[HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Internet Settings]
"ProxyEnable"=dword:00000001
"ProxyServer"="127.0.0.2:8080"
"ProxyOverride"="*.azure.com;*.azure.net;*.microsoft.com;*.windowsupdate.com;*.microsoftonline.com;*.microso
ftonline.cn;*.windows.net;*.windowsazure.com;*.windowsazure.cn;*.azure.cn;*.loganalytics.io;*.applicationins
ights.io;*.vsassets.io;*.azure-
automation.net;*.visualstudio.com,portal.office.com;*.aspnetcdn.com;*.sharepointonline.com;*.msecnd.net;*.ms
ocdn.com;*.webtrends.com"
"AutoDetect"=dword:00000000

The endpoints listed in the ProxyOverride list are limited to those endpoints needed to authenticate to Azure AD
and access Azure or Office 365 management interfaces. To extend to other cloud services, add their
administration URL to the list. This approach is designed to limit access to the wider internet to protect
privileged users from internet-based attacks. If this approach is deemed too restrictive, then consider using the
approach described below for the privileged role.

Enable Microsoft Cloud Application Security, URLs restricted list to

approved URLs (Allow most)
In our roles deployment it is recommended that for Enterprise, and Specialized deployments, where a strict deny
all web browsing is not desirable, that using the capabilities of a cloud access security broker (CASB) such as
Microsoft Defender for Cloud Apps be utilized to block access to risky, and questionable web sites. The solution
addresses a simple way to block applications and websites that have been curated. This solution is similar to
getting access to the block list from sites such as the Spamhaus Project who maintains the Domain Block List
(DBL): a good resource to use as an advanced set of rules to implement for blocking sites.
The solution will provide you:
Visibility: detect all cloud services; assign each a risk ranking; identify all users and third-party apps able to
log in
Data security: identify and control sensitive information (DLP); respond to classification labels on content
Threat protection: offer adaptive access control (AAC); provide user and entity behavior analysis (UEBA);
mitigate malware
Compliance: supply reports and dashboards to demonstrate cloud governance; assist efforts to conform to
data residency and regulatory compliance requirements
Enable Defender for Cloud Apps and connect to Defender ATP to block access the risky URLs:
In Microsoft Defender Security Center > Settings > Advanced features, set Microsoft Defender for Cloud
Apps integration > ON
In Microsoft Defender Security Center > Settings > Advanced features, set Custom network indicators > ON
In Microsoft Defender for Cloud Apps portal > Settings > Microsoft Defender ATP integration > Select Block
unsanctioned apps
 Manage local applications
The secure workstation moves to a truly hardened state when local applications are removed, including
productivity applications. Here, you add Visual Studio Code to allow connection to Azure DevOps for GitHub to
manage code repositories.
Configuring the Company Portal your for custom apps
An Intune-managed copy of the Company Portal gives you on-demand access to additional tools that you can
push down to users of the secured workstations.
In a secured mode, application installation is restricted to managed applications that are delivered by Company
Portal. However, installing the Company Portal requires access to Microsoft Store. In your secured solution, you
add and assign the Windows 10 Company Portal app for Autopilot provisioned devices.

NOTE
Make sure you assign the Company Portal app to the Secure Workstation Device Tag group used to assign the
Autopilot profile.

Deploy applications using Intune

In some situations, applications like the Microsoft Visual Studio Code are required on the secured workstation.
The following example provides instructions to install Microsoft Visual Studio Code to users in the security
group Secure Workstation Users .
Visual Studio Code is provided as an EXE package so it needs to be packaged as an .intunewin format file for
deployment using Microsoft Endpoint Manager using the Microsoft Win32 Content Prep Tool.
Download the Microsoft Win32 Content Prep Tool locally to a workstation and copy it to a directory for
packaging, for example, C:\Packages. Then create a Source and Output directory under C:\Packages.
Package Microsoft Visual Studio Code
1. Download the offline installer Visual Studio Code for Windows 64-bit.
2. Copy the downloaded Visual Studio Code exe file to C:\Packages\Source
3. Open a PowerShell console and navigate to C:\Packages
4. Type
.\IntuneWinAppUtil.exe -c C:\Packages\Source\ -s C:\Packages\Source\VSCodeUserSetup-x64-1.51.1.exe -o
C:\Packages\Output\VSCodeUserSetup-x64-1.51.1
5. Type Y to create the new output folder. The intunewin file for Visual Studio Code will be created in this
folder.
Upload VS Code to Microsoft Endpoint Manager
1. In the Microsoft Endpoint Manager admin center , browse to Apps > Windows > Add
2. Under Select app type , choose Windows app (Win32)
3. Click Select app package file , click Select a file , then select the VSCodeUserSetup-x64-1.51.1.intunewin
from C:\Packages\Output\VSCodeUserSetup-x64-1.51.1 . Click OK
4. Enter Visual Studio Code 1.51.1 in the Name field
5. Enter a description for Visual Studio Code in the Description field
6. Enter Microsoft Corporation in the Publisher Field
7. Download https://jsarray.com/images/page-icons/visual-studio-code.png and select image for the logo.
Select Next
8. Enter VSCodeSetup-x64-1.51.1.exe /SILENT in the Install command field
9. Enter C:\Program Files\Microsoft VS Code\unins000.exe in the Uninstall command field
10. Select Determine behavior based on return codes from the Device Restar t behavior dropdown list.
 Select Next
11. Select 64-bit from the Operating system architecture checkbox dropdown
12. Select Windows 10 1903 from the Minimum operating system checkbox dropdown. Select Next
13. Select Manually configure detection rules from the Rules format dropdown list
14. Click Add and then select File form the Rule type dropdown
15. Enter C:\Program Files\Microsoft VS Code in the Path field
16. Enter unins000.exe in the File or folder field
17. Select File or folder exists from the dropdown list, Select OK and then select Next
18. Select Next as there are no dependencies on this package
19. Select Add Group under Available for enrolled devices , add Privileged Users group . Click Select to
confirm group. Select Next
20. Click Create
Use PowerShell to create custom apps and settings
There are some configuration settings that we recommend, including two Defender for Endpoint
recommendations, that must be set using PowerShell. These configuration changes cannot be set via policies in
Intune.
You can also use PowerShell to extend host management capabilities. The PAW-DeviceConfig.ps1 script from
GitHub is an example script that configures the following settings:
Removes Internet Explorer
Removes PowerShell 2.0
Removes Windows Media Player
Removes Work Folders Client
Removes XPS Printing
Enables and configures Hibernate
Implements registry fix to enable AppLocker DLL rule processing
Implements registry settings for two Microsoft Defender for Endpoint recommendations that cannot be set
using Endpoint Manager.
Require users to elevate when setting a network's location
Prevent saving of network credentials
Disable Network Location Wizard - prevents users from setting network location as Private and therefore
increasing the attack surface exposed in Windows Firewall
Configures Windows Time to use NTP and sets the Auto Time service to Automatic
Downloads and sets the desktop background to a specific image to easily identify the device as a ready-to-
use, privileged workstation.
The PAW-DeviceConfig.ps1 script from GitHub.
1. Download the script [PAW-DeviceConfig.ps1] to a local device.
2. Browse to the Azure por tal > Microsoft Intune > Device configuration > PowerShell scripts > Add .
vProvide a Name for the script and specify the Script location .
3. Select Configure .
a. Set Run this script using the logged on credentials to No .
b. Select OK .
4. Select Create .
5. Select Assignments > Select groups .
a. Add the security group Secure Workstations .
b. Select Save .
 Validate and test your deployment with your first device
This enrollment assumes that you will use a physical computing device. It is recommended that as part of the
procurement process that the OEM, Reseller, distributor, or partner register devices in Windows Autopilot.
However for testing it is possible to stand up Virtual Machines as a test scenario. However note enrollment of
personally joined devices will need to be revised to allow this method of joining a client.
This method works for Virtual Machines or physical devices that have not been previously registered.
1. Start the device and wait for the username dialog to be presented
2. Press SHIFT + F10 to display command prompt
3. Type PowerShell , hit Enter
4. Type Set-ExecutionPolicy RemoteSigned , hit Enter
5. Type Install-Script GetWindowsAutopilotInfo , hit Enter
6. Type Y and click Enter to accept PATH environment change
7. Type Y and click Enter to install NuGet provider
8. Type Y to trust the repository
9. Type Run Get-WindowsAutoPilotInfo -GroupTag PAW –outputfile C:\device1.csv
10. Copy the CSV from the Virtual Machine or Physical device

Import devices into Autopilot

1. In the Microsoft Endpoint Manager admin center , go to Devices > Windows Devices >
Windows enrollment > Devices
2. Select Impor t and choose your CSV file.
3. Wait for the Group Tag to be updated to PAW and the Profile Status to change to Assigned .

NOTE
The Group Tag is used by the Secure Workstation dynamic group to make the device a member of its group,

4. Add the device to the Secure Workstations security group.

5. On the Windows 10 device you wish to configure, go to Windows Settings > Update & Security >
Recover y .
a. Choose Get star ted under Reset this PC .
b. Follow the prompts to reset and reconfigure the device with the profile and compliance policies
configured.
After you have configured the device, complete a review and check the configuration. Confirm that the first
device is configured correctly before continuing your deployment.
Assign devices
To assign devices and users, you need to map the selected profiles to your security group. All new users who
require permissions to the service must be added to the security group as well.

Using Microsoft Defender for Endpoint to monitor and respond to

security incidents
Continuously observe and monitor vulnerabilities and misconfigurations
 Utilize Microsoft Defender for Endpoint to prioritize dynamic threats in the wild
Drive correlation of vulnerabilities with endpoint detection and response (EDR) alerts
Use the dashboard to identify machine-level vulnerability during investigations
Push out remediations to Intune
Configure your Microsoft Defender Security Center. Using guidance at Threat & Vulnerability Management
dashboard overview.
Monitoring application activity using Advanced Threat Hunting
Starting at the Specialized workstation, AppLocker is enabled for monitoring of application activity on a
workstation. By default Defender for Endpoint captures AppLocker events and Advanced Hunting Queries can be
used to determine what applications, scripts, DLL files are being blocked by AppLocker.

NOTE
The Specialized and Privileged workstation profiles contain the AppLocker policies. Deployment of the policies is required
for monitoring of application activity on a client.

From the Microsoft Defender Security Center Advanced Hunting pane, use the following query to return
AppLocker events

DeviceEvents
| where Timestamp > ago(7d) and
ActionType startswith "AppControl"
| summarize Machines=dcount(DeviceName) by ActionType
| order by Machines desc

Monitoring
Understand how to review your Exposure Score
Review Security recommendation
Manage security remediations
Manage endpoint detection and response
Monitor profiles with Intune profile monitoring.

Next steps
Securing privileged access overview
Privileged access strategy
Measuring success
Security levels
Privileged access accounts
Intermediaries
Interfaces
Privileged access devices
Enterprise access model
 Security rapid modernization plan
12/13/2021 • 13 minutes to read • Edit Online

This rapid modernization plan (RAMP) will help you quickly adopt Microsoft's recommended privileged access
strategy.
This roadmap builds on the technical controls established in the privileged access deployment guidance.
Complete those steps and then use the steps in this RAMP to configure the controls for your organization.

NOTE
Many of these steps will have a green/brownfield dynamic as organizations often have security risks in the way they are
already deployed or configured accounts. This roadmap prioritizes stopping the accumulation of new security risks first,
and then later cleans up the remaining items that have already accumulated.

As you progress through the roadmap, you can utilize Microsoft Secure Score to track and compare many items
in the journey with others in similar organizations over time. Learn more about Microsoft Secure Score in the
article Secure score overview.
Each item in this RAMP is structured as an initiative that will be tracked and managed using a format that builds
on the objectives and key results (OKR) methodology. Each item includes what (objective), why, who, how, and
how to measure (key results). Some items require changes to processes and people's knoweldge/skills, while
others are simpler technology changes. Many of these initiatives will include members outside of the traditional
IT Department that should be included in the decision making and implementation of these changes to ensure
they are successfully integrated in your organization.
It is critical to work together as an organization, create partnerships, and educate people who traditionally were
not part of this process. It is critical to create and maintain buy-in across the organization, without it many
projects fail.

Separate and manage privileged accounts

Emergency access accounts
What : Ensure that you are not accidentally locked out of your Azure Active Directory (Azure AD) organization
in an emergency situation.
Why : Emergency access accounts rarely used and highly damaging to the organization if compromised, but
their availability to the organization is also critically important for the few scenarios when they are required.
 Ensure you have a plan for continuity of access that accommodates both expected and unexpected events.
Who : This initiative is typically led by Identity and Key Management and/or Security Architecture.
Sponsorship: This initiative is typically sponsored by CISO, CIO, or Director of Identity
Execution: This initiative is a collaborative effort involving
Policy and standards team document clear requirements and standards
Identity and Key Management or Central IT Operations to implement any changes
Security Compliance management monitors to ensure compliance
How : Follow the guidance in Manage emergency access accounts in Azure AD.
Measure key results:
Established Emergency access process has been designed based on Microsoft guidance that meets
organizational needs
Maintained Emergency access has been reviewed and tested within the past 90 days
Enable Azure AD Privileged Identity Management
What : Use Azure AD Privileged Identity Management (PIM) in your Azure AD production environment to
discover and secure privileged accounts
Why : Privileged Identity Management provides time-based and approval-based role activation to mitigate
the risks of excessive, unnecessary, or misused access permissions.
Who : This initiative is typically led by Identity and Key Management and/or Security Architecture.
Sponsorship: This initiative is typically sponsored by CISO, CIO, or Director of Identity
Execution: This initiative is a collaborative effort involving
Policy and standards team document clear requirements and standards (based on this
guidance)
Identity and Key Management or Central IT Operations to implement any changes
Security Compliance management monitors to ensure compliance
How : Deploy and Configure Azure AD Privileged Identity Management using the guidance in the article,
Deploy Azure AD Privileged Identity Management (PIM).
Measure key results : 100% of applicable privileged access roles are using Azure AD PIM
Identify and categorize privileged accounts (Azure AD)
What : Identify all roles and groups with high business impact that will require privileged security level
(immediately or over time). Additional specialized roles will be identified in later stages.
Why : This step is required to identify and minimize the number of people that require separate accounts
and privileged access protection
Who : This initiative is typically led by Identity and Key Management and/or Security Architecture.
Sponsorship: This initiative is typically sponsored by CISO, CIO, or Director of Identity
Execution: This initiative is a collaborative effort involving
Policy and standards team document clear requirements and standards (based on this
guidance)
Identity and Key Management or Central IT Operations to implement any changes
Security Compliance management monitors to ensure compliance
How : After turning on Azure AD Privileged Identity Management, view the users who are in the following
Azure AD roles:
Global administrator
Privileged role administrator
Exchange administrator
SharePoint administrator
 For a complete list of administrator roles, see Administrator role permissions in Azure Active Directory.

Remove any accounts that are no longer needed in those roles. Then, categorize the remaining
accounts that are assigned to admin roles:
Assigned to administrative users, but also used for non-administrative productivity purposes, like
reading and responding to email.
Assigned to administrative users and used for administrative purposes only
Shared across multiple users
For break-glass emergency access scenarios
For automated scripts
For external users

If you don't have Azure AD Privileged Identity Management in your organization, you can use the PowerShell
API. Also start with the Global Administrator role, because a Global Administrator has the same permissions
across all cloud services for which your organization has subscribed. These permissions are granted no matter
where they were assigned: in the Microsoft 365 admin center, the Azure portal, or by the Azure AD module for
Microsoft PowerShell.
Measure key results: Review and Identification of privileged access roles has been completed within the
past 90 days
Separate accounts (On-premises AD accounts)
What : Secure on-premises privileged administrative accounts, if not already done. This stage includes:
Creating separate admin accounts for users who need to conduct on-premises administrative tasks
Deploying Privileged Access Workstations for Active Directory administrators
Creating unique local admin passwords for workstations and servers
Why : Hardening the accounts used for administrative tasks. The administrator accounts should have mail
disabled and no personal Microsoft accounts should be allowed.
Who : This initiative is typically led by Identity and Key Management and/or Security Architecture.
Sponsorship: This initiative is typically sponsored by CISO, CIO, or Director of Identity
Execution: This initiative is a collaborative effort involving
Policy and standards team document clear requirements and standards (based on this
guidance)
Identity and Key Management or Central IT Operations to implement any changes
Security Compliance management monitors to ensure compliance
How : All personnel that are authorized to possess administrative privileges must have separate accounts
for administrative functions that are distinct from user accounts. Do not share these accounts
between users.
Standard user accounts - Granted standard user privileges for standard user tasks, such as email, web
browsing, and using line-of-business applications. These accounts are not granted administrative
privileges.
Administrative accounts - Separate accounts created for personnel who are assigned the appropriate
administrative privileges.
Measure key results: 100% of on-premises privileged users have separate dedicated accounts
Microsoft Defender for Identity
What : Microsoft Defender for Identity combines on-premises signals with cloud insights to monitor,
protect, and investigate events in a simplified format enabling your security teams to detect advanced
 attacks against your identity infrastructure with the ability to:
Monitor users, entity behavior, and activities with learning-based analytics
Protect user identities and credentials stored in Active Directory
Identify and investigate suspicious user activities and advanced attacks throughout the kill chain
Provide clear incident information on a simple timeline for fast triage
Why : Modern attackers may stay undetected for long periods of time. Many threats are hard to find
without a cohesive picture of your entire identity environment.
Who : This initiative is typically led by Identity and Key Management and/or Security Architecture.
Sponsorship: This initiative is typically sponsored by CISO, CIO, or Director of Identity
Execution: This initiative is a collaborative effort involving
Policy and standards team document clear requirements and standards (based on this
guidance)
Identity and Key Management or Central IT Operations to implement any changes
Security Compliance management monitors to ensure compliance
How : Deploy and enable Microsoft Defender for Identity and review any open alerts.
Measure key results : All open alerts reviewed and mitigated by the appropriate teams.

Improve credential management experience

Implement and document self-service password reset and combined security information registration
What : Enable and configure self-service password reset (SSPR) in your organization and enable the
combined security information registration experience.
Why : Users are able to reset their own passwords once they have registered. The combined security
information registration experience provides a better user experience allowing registration for Azure AD
Multi-Factor Authentication and self-service password reset. These tools when used together contribute to
lower helpdesk costs and more satisfied users.
Who : This initiative is typically led by Identity and Key Management and/or Security Architecture.
Sponsorship: This initiative is typically sponsored by CISO, CIO, or Director of Identity
Execution: This initiative is a collaborative effort involving
Policy and standards team document clear requirements and standards (based on this
guidance)
Identity and Key Management or Central IT Operations to implement any changes
Security Compliance management monitors to ensure compliance
Central IT Operations Helpdesk processes have been updated and personnel has been trained
on them
How : To enable and deploy SSPR, see the article Plan an Azure Active Directory self-service password reset
deployment.
Measure key results : Self-service password reset is fully configured and available to the organization
Protect admin accounts - Enable and require MFA / Passwordless for Azure AD privileged users
What : Require all privileged accounts in Azure AD to use strong multi-factor authentication
Why : To protect access to data and services in Microsoft 365.
Who : This initiative is typically led by Identity and Key Management and/or Security Architecture.
Sponsorship: This initiative is typically sponsored by CISO, CIO, or Director of Identity
Execution: This initiative is a collaborative effort involving
Policy and standards team document clear requirements and standards (based on this
 guidance)
Identity and Key Management or Central IT Operations to implement any changes
Security Compliance management monitors to ensure compliance
Central IT Operations Helpdesk processes have been updated and personnel has been trained
on them
Central IT Operations Service owner processes have been updated and personnel has been
trained on them
How : Turn on Azure AD Multi-Factor Authentication (MFA) and register all other highly privileged single-
user non-federated admin accounts. Require multi-factor authentication at sign-in for all individual users
who are permanently assigned to one or more of the Azure AD admin roles like:
Global administrator
Privileged Role administrator
Exchange administrator
SharePoint administrator
Require administrators to use passwordless sign-in methods such as FIDO2 security keys or Windows
Hello for Business in conjunction with unique, long, complex passwords. Enforce this change with an
organizational policy document.
Follow the guidance in the following articles, Plan an Azure AD Multi-Factor Authentication deployment and
Plan a passwordless authentication deployment in Azure Active Directory.
Measure key results : 100% of privileged users are using passwordless authentication or a strong form of
multi-factor authentication for all logons. See Privileged Access Accounts for description of multi-factor
authentication
Block legacy authentication protocols for privileged user accounts
What : Block legacy authentication protocol use for privileged user accounts.
Why : Organizations should block these legacy authentication protocols because multi-factor
authentication cannot be enforced against them. Leaving legacy authentication protocols enabled can
create an entry point for attackers. Some legacy applications may rely on these protocols and
organizations have the option to create specific exceptions for certain accounts. These exceptions should
be tracked and additional monitoring controls implemented.
Who : This initiative is typically led by Identity and Key Management and/or Security Architecture.
Sponsorship: This initiative is typically sponsored by CISO, CIO, or Director of Identity
Execution: This initiative is a collaborative effort involving
Policy and standards: establish clear requirements
Identity and Key Management or Central IT Operations Central IT Operations to implement the
policy
Security Compliance management monitors to ensure compliance
How : To block legacy authentication protocols in your organization, follow the guidance in the article
How to: Block legacy authentication to Azure AD with Conditional Access.
Measure key results :
Legacy protocols blocked: All legacy protocols are blocked for all users, with only authorized
exceptions
Exceptions are reviewed every 90 days and expire permanently within one year. Application owners
must fix all exceptions within one year of first exception approval
Application consent process
 What : Disable end-user consent to Azure AD applications.

NOTE
This change will require centralizing the decision-making process with your organization's security and identity
administration teams.

Why : Users can inadvertently create organizational risk by providing consent for an app that can maliciously
access organizational data.
Who : This initiative is typically led by Identity and Key Management and/or Security Architecture.
Sponsorship: This initiative is typically sponsored by CISO, CIO, or Director of Identity
Execution: This initiative is a collaborative effort involving
Policy and standards team document clear requirements and standards (based on this
guidance)
Identity and Key Management or Central IT Operations to implement any changes
Security Compliance management monitors to ensure compliance
Central IT Operations Helpdesk processes have been updated and personnel has been trained
on them
Central IT Operations Service owner processes have been updated and personnel has been
trained on them
How : Establish a centralized consent process to maintain centralized visibility and control of the applications
that have access to data by following the guidance in the article, Managing consent to applications and
evaluating consent requests.
Measure key results : End users are not able to consent to Azure AD application access
Clean up account and sign-in risks
What : Enable Azure AD Identity Protection and cleanup any risks that it finds.
Why : Risky user and sign-in behavior can be a source of attacks against your organization.
Who : This initiative is typically led by Identity and Key Management and/or Security Architecture.
Sponsorship: This initiative is typically sponsored by CISO, CIO, or Director of Identity
Execution: This initiative is a collaborative effort involving
Policy and standards team document clear requirements and standards (based on this
guidance)
Identity and Key Management or Central IT Operations to implement any changes
Security Compliance management monitors to ensure compliance
Central IT Operations Helpdesk processes have been updated for related support calls and
personnel has been trained on them
How : Create a process that monitors and manages user and sign-in risk. Decide if you will automate
remediation, using Azure AD Multi-Factor Authentication and SSPR, or block and require administrator
intervention.Follow the guidance in the article How To: Configure and enable risk policies.
Measure key results : The organization has zero unaddressed user and sign-in risks.

NOTE
Conditional Access policies are required to block accrual of new sign-in risks. See the Conditional access section of
Privileged Access Deployment

Admin workstations initial deployment

 What : Privileged accounts such as Global Administrators have dedicated workstations to perform
administrative tasks from.
Why : Devices where privileged administration tasks are completed are a target of attackers. Securing not
only the account but these assets are critical in reducing your attack surface area. This separation limits their
exposure to common attacks directed at productivity-related tasks like email and web browsing.
Who : This initiative is typically led by Identity and Key Management and/or Security Architecture.
Sponsorship: This initiative is typically sponsored by CISO, CIO, or Director of Identity
Execution: This initiative is a collaborative effort involving
Policy and standards team document clear requirements and standards (based on this
guidance)
Identity and Key Management or Central IT Operations to implement any changes
Security Compliance management monitors to ensure compliance
Central IT Operations Helpdesk processes have been updated and personnel has been trained
on them
Central IT Operations Service owner processes have been updated and personnel has been
trained on them
How : Initial deployment should be to the Enterprise level as described in the article Privileged Access
Deployment
Measure key results : Every privileged account has a dedicated workstation to perform sensitive tasks
from.

NOTE
This step rapidly establishes a security baseline and must be increased to specialized and privileged levels as soon as
possible.

Next steps
Securing privileged access overview
Privileged access strategy
Measuring success
Security levels
Privileged access accounts
Intermediaries
Interfaces
Privileged access devices
Enterprise access model
 Enhanced Security Admin Environment
12/13/2021 • 3 minutes to read • Edit Online

The Enhanced Security Admin Environment (ESAE) architecture (often referred to as red forest, admin forest, or
hardened forest) is an approach to provide a secure environment for Windows Server Active Directory (AD)
administrators.
Microsoft’s recommendation to use this architectural pattern has been replaced by the modern privileged access
strategy and rapid modernization plan (RAMP) guidance as the default recommended approach for securing
privileged users. The ESAE hardened administrative forest pattern (on-prem or cloud-based) is now considered
a custom configuration suitable only for exception cases listed below.

What if I already have ESAE?

For customers that have already deployed this architecture to enhance security and/or simplify multi-forest
management, there is no urgency to retire or replace an ESAE implementation if it's being operated as designed
and intended. As with any enterprise systems, you should maintain the software in it by applying security
updates and ensuring software is within support lifecycle.
Microsoft also recommends organizations with ESAE / hardened forests adopt the modern privileged access
strategy using the rapid modernization plan (RAMP) guidance. This complements an existing ESAE
implementation and provides appropriate security for roles not already protected by ESAE including Azure AD
Global Administrators, sensitive business users, and standard enterprise users. For more information, see the
article Securing privileged access security levels.

Why change the recommendation?

When ESAE was originally designed 10 years ago, the focus was on on-premise environments with AD as the
local identity provider. ESAE / hardened forest implementations focus on protecting Windows Server Active
Directory administrators.
Microsoft recommends the new cloud-based solutions because they can be deployed more quickly to protect a
broader scope of administrative and business-sensitive roles and systems.
The privileged access strategy provides protections and monitoring for a much larger set of sensitive users,
while providing incremental lower-cost steps to rapidly build security assurances.
While still valid for specific use cases, ESAE hardened forest implementations are more costly and more difficult
to use, requiring more operational support compared to the newer cloud-based solution (due to the complex
nature of that architecture). ESAE implementations are designed to protect only Windows Server Active
Directory administrators. The cloud based privileged access strategy provides protections and monitoring for a
much larger set of sensitive users, while providing incremental lower-cost steps to rapidly build security
assurances.

What are the valid ESAE use cases?

While not a mainstream recommendation, this architectural pattern is valid in a limited set of scenarios.
In these exception cases, the organization must accept the increased technical complexity and operational costs
of the solution. The organization must have a sophisticated security program to measure risk, monitor risk, and
apply consistent operational rigor to the usage and maintenance of the ESAE implementation.
Example scenarios include:
Isolated on-premises environments - where cloud services are unavailable such as offline research
laboratories, critical infrastructure or utilities, disconnected operational technology (OT) environments such
as Supervisory control and data acquisition (SCADA) / Industrial Control Systems (ICS), and public sector
customers that are fully reliant on on-premises technology.
Highly regulated environments – industry or government regulation may specifically require an
administrative forest configuration.
High level security assurance is mandated - organizations with low risk tolerance that are willing to
accept the increased complexity and operational cost of the solution.

NOTE
While Microsoft no longer recommends an isolated hardened forest model for most scenarios at most organizations,
Microsoft still operates a similar architecture internally (and associated support processes and personnel) because of the
extreme security requirements for providing trusted cloud services to organizations around the globe.

Next steps
Review the privileged access strategy and rapid modernization plan (RAMP) guidance for providing secure
environments for privileged users.
 Microsoft Security Best Practices module: Privileged
administration
12/13/2021 • 2 minutes to read • Edit Online

Administrative accounts with privileged access to the environment (and associated elements like groups and
workstations) must be protected at the highest levels of security assurances to ensure all other security
assurances aren't undermined.
See the Administration topic for more information.
The following videos provide guidance on administration. You can also download the PowerPoint slides
associated with these videos.

Part 1: Introduction (05:40)

Part 2: Admin Quantity (03:14)

Part 3: Managed and Separate Admin Accounts (03:38)

Part 4: Emergency Access (02:28)

Part 5: Containing Attack Pivot Risk (02:42)

Part 6: Admin Account Protection (05:25)

Part 7: Admin Workstation Security (04:09)

Part 8: Enforcing Access Security (03:13)

Part 9: Simplify Permissions (03:31)

Part 10: Admin Account Lifecycle (02:53)

Next steps
For additional security guidance from Microsoft, see Microsoft security documentation.
 Administration
12/13/2021 • 9 minutes to read • Edit Online

Administration is the practice of monitoring, maintaining, and operating Information Technology (IT) systems to
meet service levels that the business requires. Administration introduces some of the highest impact security
risks because performing these tasks requires privileged access to a very broad set of these systems and
applications. Attackers know that gaining access to an account with administrative privileges can get them
access to most or all of the data they would target, making the security of administration one of the most critical
security areas.
As an example, Microsoft makes significant investments in protection and training of administrators for our
cloud systems and IT systems:

Microsoft’s recommended core strategy for administrative privileges is to use the available controls to reduce
risk
Reduce risk exposure (scope and time) – The principle of least privilege is best accomplished with modern
controls that provide privileges on demand. This help to limit risk by limiting administrative privileges exposure
by:
Scope – Just Enough Access (JEA) provides only the required privileges for the administrative operation
required (vs. having direct and immediate privileges to many or all systems at a time, which is almost
never required).
Time – Just in Time (JIT) approaches provided the required privileged as they are needed.
Mitigate the remaining risks – Use a combination of preventive and detective controls to reduce risks
such as isolating administrator accounts from the most common risks phishing and general web
browsing, simplifying and optimizing their workflow, increasing assurance of authentication decisions,
and identifying anomalies from normal baseline behavior that can be blocked or investigated.
Microsoft has captured and documented best practices for protecting administrative accounts and published
prioritized roadmaps for protecting privileged access that can be used as references for prioritizing mitigations
for accounts with privileged access.
Securing Privileged Access (SPA) roadmap for administrators of on premises Active Directory
Guidance for securing administrators of Azure Active Directory

Minimize number of critical impact admins

Grant the fewest number of accounts to privileges that can have a critical business impact
Each admin account represents potential attack surface that an attacker can target, so minimizing the number of
accounts with that privilege helps limit the overall organizational risk. Experience has taught us that
membership of these privileged groups grows naturally over time as people change roles if membership not
actively limited and managed.
We recommend an approach that reduces this attack surface risk while ensuring business continuity in case
something happens to an administrator:
Assign at least two accounts to the privileged group for business continuity
When two or more accounts are required, provide justification for each member including the original
two
Regularly review membership & justification for each group member

Managed accounts for admins

Ensure all critical impact admins in are managed by enterprise directory to follow organizational policy
enforcement.
Consumer accounts such as Microsoft accounts like @Hotmail.com, @live.com, @outlook.com, don’t offer
sufficient security visibility and control to ensure the organization’s policies and any regulatory requirements
are being followed. Because Azure deployments often start small and informally before growing into enterprise-
managed tenants, some consumer accounts remain as administrative accounts long afterward for example,
original Azure project managers, creating blind spots, and potential risks.

Separate accounts for admins

Ensure all critical impact admins have a separate account for administrative tasks (vs the account they use for
email, web browsing, and other productivity tasks).
Phishing and web browser attacks represent the most common attack vectors to compromise accounts,
including administrative accounts.
Create a separate administrative account for all users that have a role requiring critical privileges. For these
administrative accounts, block productivity tools like Office 365 email (remove license). If possible, block
arbitrary web browsing (with proxy and/or application controls) while allowing exceptions for browsing to the
Azure portal and other sites required for administrative tasks.

No standing access / Just in Time privileges

Avoid providing permanent “standing” access for any critical impact accounts
Permanent privileges increase business risk by increasing the time an attacker can use the account to do
damage. Temporary privileges force attackers targeting an account to either work within the limited times the
admin is already using the account or to initiate privilege elevation (which increases their chance of being
detected and removed from the environment).
Grant privileges required only as required using one of these methods:
Just in Time - Enable Azure AD Privileged Identity Management (PIM) or a third party solution to
require following an approval workflow to obtain privileges for critical impact accounts
Break glass – For rarely used accounts, follow an emergency access process to gain access to the
accounts. This is preferred for privileges that have little need for regular operational usage like members
of global admin accounts.

Emergency access or ‘Break Glass’ accounts

Ensure you have a mechanism for obtaining administrative access in case of an emergency
While rare, sometimes extreme circumstances arise where all normal means of administrative access are
unavailable.
We recommend following the instructions at Managing emergency access administrative accounts in Azure AD
and ensure that security operations monitor these accounts carefully.

Admin workstation security

Ensure critical impact admins use a workstation with elevated security protections and monitoring
Attack vectors that use browsing and email like phishing are cheap and common. Isolating critical impact
admins from these risks will significantly lower your risk of a major incident where one of these accounts is
compromised and used to materially damage your business or mission.
Choose level of admin workstation security based on the options available at https://aka.ms/securedworkstation
Highly Secure Productivity Device (Enhanced Security Workstation or Specialized
Workstation)
You can start this security journey for critical impact admins by providing them with a higher security
workstation that still allows for general browsing and productivity tasks. Using this as an interim step
helps ease the transition to fully isolated workstations for both the critical impact admins as well as the IT
staff supporting these users and their workstations.
Privileged Access Workstation (Specialized Workstation or Secured Workstation)
These configurations represent the ideal security state for critical impact admins as they heavily restrict
access to phishing, browser, and productivity application attack vectors. These workstations don’t allow
general internet browsing, only allow browser access to Azure portal and other administrative sites.

Critical impact admin dependencies – Account/Workstation

Carefully choose the on-premises security dependencies for critical impact accounts and their workstations
To contain the risk from a major incident on-premises spilling over to become a major compromise of cloud
assets, you must eliminate or minimize the means of control that on premises resources have to critical impact
accounts in the cloud. As an example, attackers who compromise the on premises Active Directory can access
and compromise cloud-based assets that rely on those accounts like resources in Azure, Amazon Web Services
(AWS), ServiceNow, and so on. Attackers can also use workstations joined to those on premises domains to gain
access to accounts and services managed from them.
Choose the level of isolation from on premises means of control also known as security dependencies for critical
impact accounts
User Accounts – Choose where to host the critical impact accounts
Native Azure AD Accounts -*Create Native Azure AD Accounts that are not synchronized with on-
 premises active directory
Synchronize from On Premises Active Directory (Not Recommended see Don’t synchronize on-
premises admin accounts to cloud identity providers)- Leverage existing accounts hosted in the on
premises active directory.
Workstations – Choose how you will manage and secure the workstations used by critical admin
accounts:
Native Cloud Management & Security (Recommended) - Join workstations to Azure AD &
Manage/Patch them with Intune or other cloud services. Protect and Monitor with Windows
Microsoft Defender ATP or another cloud service not managed by on premises based accounts.
Manage with Existing Systems - Join existing AD domain & leverage existing
management/security.
This is related to the Don’t synchronize on-premises admin accounts to cloud identity providers to cloud identity
providers guidance in the administration section that mitigates the inverse risk of pivoting from cloud assets to
on-premises assets

Passwordless Or multi-factor authentication for admins

Require all critical impact admins to use passwordless authentication or multi-factor authentication (MFA).
Attack methods have evolved to the point where passwords alone cannot reliably protect an account. This is well
documented in a Microsoft Ignite Session.
Administrative accounts and all critical accounts should use one of the following methods of authentication.
These capabilities are listed in preference order by highest cost/difficulty to attack (strongest/preferred options)
to lowest cost/difficult to attack:
Passwordless (such as Windows Hello)
https://aka.ms/HelloForBusiness
Passwordless (Authenticator App)
/azure/active-directory/authentication/howto-authentication-phone-sign-in
Multifactor Authentication
/azure/active-directory/authentication/howto-mfa-userstates
Note that SMS Text Message based MFA has become very inexpensive for attackers to bypass, so we
recommend you avoid relying on it. This option is still stronger than passwords alone, but is much weaker than
other MFA options

Enforce conditional access for admins - Zero Trust

Authentication for all admins and other critical impact accounts should include measurement and enforcement
of key security attributes to support a Zero Trust strategy.
Attackers compromising Azure Admin accounts can cause significant harm. Conditional Access can significantly
reduce that risk by enforcing security hygiene before allowing access to Azure management.
Configure Conditional Access policy for Azure management that meets your organization’s risk appetite and
operational needs.
Require Multifactor Authentication and/or connection from designated work network
Require Device integrity with Microsoft Defender ATP (Strong Assurance)
Avoid granular and custom permissions
Avoid permissions that specifically reference individual resources or users
Specific permissions create unneeded complexity and confusion as they don’t carry the intention to new similar
resources. This then accumulates into a complex legacy configuration that is difficult to maintain or change
without fear of “breaking something” – negatively impacting both security and solution agility.
Instead of assigning specific resource-specific permissions, use either
Management Groups for enterprise-wide permissions
Resource groups for permissions within subscriptions
Instead of granting permissions to specific users, assign access to groups in Azure AD. If there isn’t an
appropriate group, work with the identity team to create one. This allows you to add and remove group
members externally to Azure and ensure permissions are current, while also allowing the group to be used for
other purposes such as mailing lists.

Use built-in roles

Use built-in roles for assigning permissions where possible.
Customization leads to complexity that increases confusion and makes automation more complex, challenging,
and fragile. These factors all negatively impact security
We recommend that you evaluate the built-in roles designed to cover most normal scenarios. Custom roles are
a powerful and sometimes useful capability, but they should be reserved for cases when built in roles won’t
work.

Establish lifecycle management for critical impact accounts

Ensure you have a process for disabling or deleting administrative accounts when admin personnel leave the
organization (or leave administrative positions)
See Regularly review critical access for more details.

Attack simulation for critical impact accounts

Regularly simulate attacks against administrative users with current attack techniques to educate and empower
them.
People are a critical part of your defense, especially your personnel with access to critical impact accounts.
Ensuring these users (and ideally all users) have the knowledge and skills to avoid and resist attacks will reduce
your overall organizational risk.
You can use Office 365 Attack Simulation capabilities or any number of third party offerings.

Next steps
For additional security guidance from Microsoft, see Microsoft security documentation.
 Human-operated ransomware
12/13/2021 • 3 minutes to read • Edit Online

Human-operated ransomware is a large and growing attack trend that represents a threat to organizations in
every industry.
Human-operated ransomware is different than commodity ransomware. These “hands-on-keyboard” attacks
target an organization rather than a single device and leverage human attackers’ knowledge of common system
and security misconfigurations to infiltrate the organization, navigate the enterprise network, and adapt to the
environment and its weaknesses as they go.
Hallmarks of these human-operated ransomware attacks typically include credential theft and lateral movement
and can result in deployment of a ransomware payload to high business impact resources the attackers choose.
These attacks can be catastrophic to business operations and are difficult to clean up, requiring complete
adversary eviction to protect against future attacks. Unlike commodity ransomware that only requires malware
remediation, human-operated ransomware will continue to threaten your business operations after the initial
encounter.
This figure shows how this extortion-based attack that uses maintenance and security configuration gaps and
privileged access is growing in impact and likelihood.

Protect your organization against ransomware and extortion

For a comprehensive view of ransomware and extortion and how to protect your organization, use the
information in the Human-Operated Ransomware Mitigation Project Plan PowerPoint presentation.
Here's a summary of the guidance:
 The stakes of ransomware and extortion are high.
However, the attacks have weaknesses that can mitigate your likelihood of being attacked.
There are three phases to configure your infrastructure to exploit attack weaknesses.
For the three phases to exploit attack weaknesses, see the Protect your organization against ransomware and
extortion solution to quickly configure your IT infrastructure for the best protection:
1. Prepare your organization so you can recover from an attack without having to pay the ransom.
2. Limit the scope of damage of an attack by protecting privileged roles.
3. Make it harder for a ransomware attacker to get into your environment by incrementally removing risks.

You can also see an overview of the phases as levels of protection against ransomware attackers with the
Protect your organization from ransomware poster.

Additional ransomware resources

Key information from Microsoft:
The growing threat of ransomware, Microsoft On the Issues blog post on July 20, 2021
Rapidly protect against ransomware and extortion
2021 Microsoft Digital Defense Report (see pages 10-19)
Ransomware: A pervasive and ongoing threat threat analytics report in the Microsoft 365 Defender portal
Microsoft’s DART ransomware approach and best practices
Microsoft 365:
 Deploy ransomware protection for your Microsoft 365 tenant
Maximize Ransomware Resiliency with Azure and Microsoft 365
Recover from a ransomware attack
Malware and ransomware protection
Protect your Windows 10 PC from ransomware
Handling ransomware in SharePoint Online
Threat analytics reports for ransomware in the Microsoft 365 Defender portal
Microsoft 365 Defender:
Find ransomware with advanced hunting
Microsoft Defender for Cloud Apps:
Create anomaly detection policies in Defender for Cloud Apps
Microsoft Azure:
Azure Defenses for Ransomware Attack
Maximize Ransomware Resiliency with Azure and Microsoft 365
Backup and restore plan to protect against ransomware
Help protect from ransomware with Microsoft Azure Backup (26 minute video)
Recovering from systemic identity compromise
Advanced multistage attack detection in Microsoft Sentinel
Fusion Detection for Ransomware in Microsoft Sentinel
Microsoft Security team blog posts:
3 steps to prevent and recover from ransomware (September 2021)
A guide to combatting human-operated ransomware: Part 1 (September 2021)
Key steps on how Microsoft's Detection and Response Team (DART) conducts ransomware incident
investigations.
A guide to combatting human-operated ransomware: Part 2 (September 2021)
Recommendations and best practices.
Becoming resilient by understanding cybersecurity risks: Part 4—navigating current threats (May 2021)
See the Ransomware section.
Human-operated ransomware attacks: A preventable disaster (March 2020)
Includes attack chain analyses of actual attacks.
Ransomware response—to pay or not to pay? (December 2019)
Norsk Hydro responds to ransomware attack with transparency (December 2019)
 Rapidly protect against ransomware and extortion
12/13/2021 • 6 minutes to read • Edit Online

NOTE
This guidance will be updated as new information becomes available.

Mitigating ransomware and extortion attacks is an urgent priority for organizations because of the high impact
of these attacks and high likelihood an organization will experience one.
Ransomware is a type of extortion attack that encrypts files and folders, preventing access to important data.
Criminals use ransomware to extort money from victims by demanding money, usually in form of
cryptocurrencies, in exchange for a decryption key. Criminals also often use ransomware to extort money from
victims in exchange for not releasing sensitive data to the dark web or the public internet.
While early ransomware mostly used malware that spread with phishing or between devices, human-operated
ransomware has emerged where a gang of active attackers, driven by human intelligence, target an organization
rather than a single device or set of devices and leverage the attackers’ knowledge of common system and
security misconfigurations and vulnerabilities to infiltrate the organization, navigate the enterprise network, and
adapt to the environment and its weaknesses as they go.
These attacks can be catastrophic to business operations and are difficult to clean up, requiring complete
adversary eviction to protect against future attacks. Unlike early forms of ransomware that only required
malware remediation, human-operated ransomware can continue to threaten your business operations after the
initial encounter.

Sophistication of ransomware and extortion attacks

Technical hallmarks of these human-operated ransomware attacks typically include credential theft and lateral
movement and can result in deployment of ransomware payloads to many high-value resources in order to
encourage payment of the ransom. The attacker model is often effective and highly tuned, including targeting
destruction or encryption of backups, network and enterprise documentation, and other artifacts that would let
the organization rebuild without paying the ransom.
The attacks also have sophisticated business models, with attackers setting their ransom prices based on
internal financial documentation from the company, cyber-insurance coverage levels, and typical regulatory
compliance fines the company would have to pay.

Organization of the guidance in this solution

This guidance provides concrete instructions on how to best prepare your organization from many forms of
ransomware and extortion.
To help you understand how to protect your organization from ransomware, this guidance is organized into
prioritized phases. Each phase corresponds to a separate article. The priority order is designed to ensure you
reduce risk as fast as possible with each phase, building on an assumption of extreme urgency that will override
normal security and IT priorities to avoid or mitigate these devastating attacks.
It is vital to note that this guidance is structured as prioritized phases that you should follow in the prescribed
order. To best adapt this guidance to your situation:
1. Stick with the recommended priorities
Use the phases as a starting plan for what to do first, next, and later so you get the most impactful
elements first. These recommendations have been prioritized using the Zero Trust principle of assume
breach, which focuses on minimizing business risk by assuming the attackers can successfully gain access
to your environment through one or more methods.
2. Be proactive and flexible (but don’t skip impor tant tasks)
Scan through the implementation checklists for all sections of all three phases to see if there are any
areas and tasks that you can quickly complete earlier (e.g., already have access to a cloud service that
hasn’t been utilized but could be quickly and easily configured). As you look over the whole plan, be very
careful that these later areas and tasks don’t delay completion of critically important areas like backups
and privileged access!
3. Do some items in parallel
Trying to do everything at once can be overwhelming, but some items can naturally be done in parallel.
Staff on different teams can be working on tasks at the same time (e.g., backup team, endpoint team,
identity team), while also driving for completion of the phases in priority order.
The items in the implementation checklists are in the recommended order of prioritization, not a technical
dependency order. Use the checklists to confirm and modify your existing configuration as needed and in a way
that works within your organization. For example, in the most important backup element, you backup some
systems, but they may not be offline/immutable, or you may not test the full enterprise restore procedures, or
you may not have backups of critical business systems or critical IT systems like Active Directory Domain
Services (AD DS) domain controllers.

NOTE
See the 3 steps to prevent and recover from ransomware (September 2021) Microsoft security blog post for an additional
summary of this process.

Phase 1. Prepare your recovery plan

This phase is designed to minimize the monetary incentive from ransomware attackers by making it:
Much harder to access and disrupt systems or encrypt or damage key organization data.
Easier for your organization to recover from an attack without paying the ransom.
 NOTE
While restoring many or all enterprise systems is a difficult endeavor, the alternative of paying an attacker for a recovery
key and then using tools written by the attackers to recover systems and data is a much worse option.

Phase 2. Limit the scope of damage

Make the attackers work a lot harder to gain access to multiple business critical systems via privileged access
roles. Limiting the attacker’s ability to get privileged access makes it much harder to profit off of an attack on
your organization, making it more likely they will give up and go elsewhere.
Phase 3. Make it hard to get in
This last set of tasks is important to raise friction for entry but will take time to complete as part of a larger
security journey. The goal of this phase is for attackers to have to work a lot harder to obtain access to your on-
premises or cloud infrastructures at the various common points of entry for attacks. There are a lot of these
tasks, so it’s important to prioritize your work here based on how fast you can accomplish these with your
current resources.
While many of these will be familiar and/or easy to quickly accomplish, it’s critically important that your work
on phase 3 should not slow down your progress on phases 1 and 2!

At a glance
You can also see an overview of the phases and their implementation checklists as levels of protection against
ransomware attackers with the Protect your organization from ransomware poster.

Next step

Start with Phase 1 to prepare your organization to recover from an attack without having to pay the ransom.

Additional ransomware resources

Key information from Microsoft:
The growing threat of ransomware, Microsoft On the Issues blog post on July 20, 2021
Human-operated ransomware
2021 Microsoft Digital Defense Report (see pages 10-19)
Ransomware: A pervasive and ongoing threat threat analytics report in the Microsoft 365 Defender portal
Microsoft’s DART ransomware approach and best practices
Microsoft 365:
 Deploy ransomware protection for your Microsoft 365 tenant
Maximize Ransomware Resiliency with Azure and Microsoft 365
Recover from a ransomware attack
Malware and ransomware protection
Protect your Windows 10 PC from ransomware
Handling ransomware in SharePoint Online
Threat analytics reports for ransomware in the Microsoft 365 Defender portal
Microsoft 365 Defender:
Find ransomware with advanced hunting
Microsoft Azure:
Azure Defenses for Ransomware Attack
Maximize Ransomware Resiliency with Azure and Microsoft 365
Backup and restore plan to protect against ransomware
Help protect from ransomware with Microsoft Azure Backup (26 minute video)
Recovering from systemic identity compromise
Advanced multistage attack detection in Microsoft Sentinel
Fusion Detection for Ransomware in Microsoft Sentinel
Microsoft Defender for Cloud Apps:
Create anomaly detection policies in Defender for Cloud Apps
Microsoft Security team blog posts:
3 steps to prevent and recover from ransomware (September 2021)
A guide to combatting human-operated ransomware: Part 1 (September 2021)
Key steps on how Microsoft's Detection and Response Team (DART) conducts ransomware incident
investigations.
A guide to combatting human-operated ransomware: Part 2 (September 2021)
Recommendations and best practices.
Becoming resilient by understanding cybersecurity risks: Part 4—navigating current threats (May 2021)
See the Ransomware section.
Human-operated ransomware attacks: A preventable disaster (March 2020)
Includes attack chain analyses of actual attacks.
Ransomware response—to pay or not to pay? (December 2019)
Norsk Hydro responds to ransomware attack with transparency (December 2019)
 Phase 1: Prepare your recovery plan
12/13/2021 • 6 minutes to read • Edit Online

The first thing you should do for these attacks is prepare your organization so that it has a viable alternative to
paying the ransom. While attackers in control of your organization have a variety of ways to pressure you into
paying, the demands primarily focus on two categories:
Pay to regain access
Attackers demand payment under the threat that they won’t give you back access to your systems and
data. This is most frequently done by encrypting your systems and data and demanding payment for the
decryption key.

IMPORTANT
Paying the ransom isn’t as simple and clean of a solution as it may seem. Because you're dealing with criminals
that are only motivated by payment (and often relatively amateur operators who are using a toolkit provided by
someone else), there is a lot of uncertainty around how well paying the ransom will actually work. There is no legal
guarantee that they will provide a key that decrypts 100% of your systems and data, or even provide a key at all.
The process to decrypt these systems uses homegrown attacker tools, which is often a clumsy and manual
process.

Pay to avoid disclosure

Attackers demand payment in exchange for not releasing sensitive or embarrassing data to the dark web
(other criminals) or the general public.
To avoid being forced into payment (the profitable situation for attackers), the most immediate and effective
action you can take is to ensure your organization can restore your entire enterprise from immutable storage,
which neither the attacker nor you can modify.
Identifying the most sensitive assets and protecting them at a higher level of assurance is also critically
important but is a longer and more challenging process to execute. We don’t want you to hold up other areas in
phases 1 or 2, but we recommend you get the process started by bringing together business, IT, and security
stakeholders to ask and answer questions like:
What business assets would be the most damaging if compromised? For example, what assets would
business leadership would be willing to pay an extortion demand if attackers controlled them?
How do these business assets translate to IT assets (files, applications, databases, servers, etc.)?
How can we protect or isolate these assets so that attackers with access to the general IT environment can’t
access them?

Secure backups
You must ensure that critical systems and their data are backed up and backups are protected against deliberate
erasure or encryption by an attacker.
Attacks on your backups focus on crippling your organization’s ability to respond without paying, frequently
targeting backups and key documentation required for recovery to force you into paying extortion demands.
Most organizations don’t protect backup and restoration procedures against this level of intentional targeting.
 NOTE
This preparation also improves resilience to natural disasters and rapid attacks like WannaCry and (Not)Petya.

Backup and restore plan to protect against ransomware addresses what to do before an attack to protect your
critical business systems and during an attack to ensure a rapid recovery of your business operations.
Program and project member accountabilities
This table describes the overall protection of your data from ransomware in terms of a sponsorship/program
management/project management hierarchy to determine and drive results.

L EA D IM P L EM EN TO R A C C O UN TA B IL IT Y

Central IT Operations or CIO Executive sponsorship

Program lead from Central IT Drive results and cross-team

infrastructure collaboration

Central IT Infrastructure/Backup Enable Infrastructure backup

Central IT Productivity / End User Enable OneDrive Backup

Security Architecture Advise on configuration and standards

Security Policy and Standards Update standards and policy

documents

Security Compliance Management Monitor to ensure compliance

Implementation checklist
Apply these best practices to secure your backup infrastructure.

DO N E TA SK DESC RIP T IO N

Backup all critical systems Allows you to recover data up to the

automatically on a regular schedule. last backup.

Regularly exercise your business Ensures rapid recovery of business

continuity/disaster recovery (BC/DR) operations by treating a ransomware
plan. or extortion attack with the same
importance as a natural disaster.

Protect backups against deliberate Backups that are accessible by

erasure and encryption: attackers can be rendered unusable for
business recovery.
- Strong Protection – Require out of
band steps (MFA or PIN) before
modifying online backups (such as
Azure Backup).

- Strongest Protection – Store backups

in online immutable storage (such as
Azure Blob) and/or fully offline or off-
site.
 DO N E TA SK DESC RIP T IO N

Protect supporting documents Attackers deliberately target these

required for recovery such as resources because it impacts your
restoration procedure documents, ability to recover.
your configuration management
database (CMDB), and network
diagrams.

Implementation results and timelines

Within 30 days, ensure that Mean Time to Recover (MTTR) meets your BC/DR goal, as measured during
simulations and real-world operations.

Data protection
You must implement data protection to ensure rapid and reliable recovery from a ransomware attack and to
block some techniques of attackers.
Ransomware extortion and destructive attacks only work when all legitimate access to data and systems is lost.
Ensuring that attackers cannot remove your ability to resume operations without payment will protect your
business and undermine the monetary incentive for attacking your organization.
Program and project member accountabilities
This table describes the overall protection of your organization data from ransomware in terms of a
sponsorship/program management/project management hierarchy to determine and drive results.

L EA D IM P L EM EN TO R A C C O UN TA B IL IT Y

Central IT Operations or CIO Executive sponsorship

Program lead from Data Security Drive results and cross-team

collaboration

Central IT Productivity / End User Implement changes to Microsoft 365

tenant for OneDrive and Protected
Folders

Central IT Infrastructure/Backup Enable Infrastructure backup

Business / Application Identify critical business assets

Security Architecture Advise on configuration and standards

Security Policy and Standards Update standards and policy

documents

Security Compliance Management Monitor to ensure compliance

User Education Team Ensure guidance for users reflects

policy updates

Implementation checklist
Apply these best practices to protect your organization data.
 DO N E TA SK DESC RIP T IO N

Migrate your organization to the User data in the Microsoft cloud can
cloud: be protected by built-in security and
data management features.
- Move user data to cloud solutions
like OneDrive/SharePoint to take
advantage of versioning and recycle
bin capabilities.

- Educate users on how to recover

their files by themselves to reduce
delays and cost of recovery.

Designate Protected Folders. Makes it more difficult for

unauthorized applications to modify
the data in these folders.

Review your permissions: Reduces risk from broad access-

enabling ransomware activities.
- Discover broad write/delete
permissions on file shares, SharePoint,
and other solutions. Broad is defined
as many users having write/delete
permissions for business-critical data.

- Reduce broad permissions while

meeting business collaboration
requirements.

- Audit and monitor to ensure broad

permissions don’t reappear.

Next step

Continue with Phase 2 to limit the scope of damage of an attack by protecting privileged roles.

Additional ransomware resources

Key information from Microsoft:
The growing threat of ransomware, Microsoft On the Issues blog post on July 20, 2021
Human-operated ransomware
Rapidly protect against ransomware and extortion
2021 Microsoft Digital Defense Report (see pages 10-19)
 Ransomware: A pervasive and ongoing threat threat analytics report in the Microsoft 365 Defender portal
Microsoft’s DART ransomware approach and best practices
Microsoft 365:
Deploy ransomware protection for your Microsoft 365 tenant
Maximize Ransomware Resiliency with Azure and Microsoft 365
Recover from a ransomware attack
Malware and ransomware protection
Protect your Windows 10 PC from ransomware
Handling ransomware in SharePoint Online
Threat analytics reports for ransomware in the Microsoft 365 Defender portal
Microsoft 365 Defender:
Find ransomware with advanced hunting
Microsoft Azure:
Azure Defenses for Ransomware Attack
Maximize Ransomware Resiliency with Azure and Microsoft 365
Backup and restore plan to protect against ransomware
Help protect from ransomware with Microsoft Azure Backup (26 minute video)
Recovering from systemic identity compromise
Advanced multistage attack detection in Microsoft Sentinel
Fusion Detection for Ransomware in Microsoft Sentinel
Microsoft Defender for Cloud Apps:
Create anomaly detection policies in Defender for Cloud Apps
Microsoft Security team blog posts:
A guide to combatting human-operated ransomware: Part 1 (September 2021)
Key steps on how Microsoft's Detection and Response Team (DART) conducts ransomware incident
investigations.
A guide to combatting human-operated ransomware: Part 2 (September 2021)
Recommendations and best practices.
3 steps to prevent and recover from ransomware (September 2021)
Becoming resilient by understanding cybersecurity risks: Part 4—navigating current threats (May 2021)
See the Ransomware section.
Human-operated ransomware attacks: A preventable disaster (March 2020)
Includes attack chain analyses of actual attacks.
Ransomware response—to pay or not to pay? (December 2019)
Norsk Hydro responds to ransomware attack with transparency (December 2019)
 Phase 2: Limit the scope of damage
12/13/2021 • 4 minutes to read • Edit Online

In this phase, you prevent attackers from obtaining a large scope of access for potential damage to data and
systems by protecting privileged roles.

Privileged access strategy

You must implement a comprehensive strategy to reduce the risk of privileged access compromise.
All other security controls can easily be invalidated by an attacker with privileged access in your environment.
Ransomware attackers use privileged access as a quick path to control all critical assets in the organization for
their extortion.
Program and project member accountabilities
This table describes a privileged access strategy against ransomware in terms of a sponsorship/program
management/project management hierarchy to determine and drive results.

L EA D IM P L EM EN TO R A C C O UN TA B IL IT Y

CISO or CIO Executive sponsorship

Program lead Drive results and cross-team

collaboration

IT and Security Architects Prioritize components integrate into

architectures

Identity and Key Management Implement identity changes

Central IT Productivity / End User Team Implement changes to devices and

Office 365 tenant

Security Policy and Standards Update standards and policy

documents

Security Compliance Management Monitor to ensure compliance

User Education Team Update any password guidance

Implementation checklist
Build a multi-part strategy using the guidance at https://aka.ms/SPA that includes this checklist.

DO N E TA SK DESC RIP T IO N

Enforce end-to-end session security. Explicitly validates the trust of users

and devices before allowing access to
administrative interfaces (using Azure
AD Conditional Access).
 DO N E TA SK DESC RIP T IO N

Protect and monitor identity systems. Prevents privilege escalation attacks

including directories, identity
management, administrator accounts
and groups, and consent grant
configuration.

Mitigate lateral traversal. Ensures that compromising a single

device does not immediately lead to
control of many or all other devices
using local account passwords, service
account passwords, or other secrets.

Ensure rapid threat response. Limits an adversary's access and time

in the environment. See Detection and
Response for more information.

Implementation results and timelines

Try to achieve these results in 30-90 days:
100 % of admins required to use secure workstations
100 % local workstation/server passwords randomized
100 % deployment of privilege escalation mitigations

Detection and response

Your organization needs responsive detection and remediation of common attacks on endpoints, email, and
identities. Minutes matter. You must rapidly remediate common attack entry points to limit the attacker’s time to
laterally traverse your organization.
Program and project member accountabilities
This table describes the improvement of your detection and response capability against ransomware in terms of
a sponsorship/program management/project management hierarchy to determine and drive results.

L EA D IM P L EM EN TO R A C C O UN TA B IL IT Y

CISO or CIO Executive sponsorship

Program lead from Security Drive results and cross-team

Operations collaboration

Central IT Infrastructure Team Implement client and server

agents/features

Security Operations Integrate any new tools into security

operations processes

Central IT Productivity / End User Team Enable features for Defender for
Endpoint, Defender for Office 365,
Defender for Identity, and Defender for
Cloud Apps
 L EA D IM P L EM EN TO R A C C O UN TA B IL IT Y

Central IT Identity Team Implement Azure AD security and

Defender for Identity

Security Architects Advise on configuration, standards,

and tooling

Security Policy and Standards Update standards and policy

documents

Security Compliance Management Monitor to ensure compliance

Implementation checklist
Apply these best practices for improving your detection and response.

DO N E TA SK DESC RIP T IO N

Prioritize common entry points: Ransomware (and other) operators

favor endpoint, email, identity, and
- Use integrated Extended Detection RDP as entry points.
and Response (XDR) tools like
Microsoft 365 Defender to provide
high quality alerts and minimize
friction and manual steps during
response.

- Monitor for brute-force attempts like

password spray.

Monitor for an adversary disabling Attackers target security detection

security (this is often part of an attack facilities to more safely continue their
chain), such as: attack.

- Event log clearing, especially the

Security Event log and PowerShell
Operational logs.

- Disabling of security tools and

controls (associated with some
groups).

Don’t ignore commodity malware. Ransomware attackers regularly

purchase access to target
organizations from dark markets.

Integrate outside experts into Experience counts for detection and

processes to supplement expertise, recovery.
such as the Microsoft Detection and
Response Team (DART).

Rapidly isolate compromised Windows 10 integration makes this

computers using Defender for easy.
Endpoint.
Next step

Continue with Phase 3 to make it hard for an attacker to get into your environment by incrementally removing
risks.

Additional ransomware resources

Key information from Microsoft:
The growing threat of ransomware, Microsoft On the Issues blog post on July 20, 2021
Human-operated ransomware
Rapidly protect against ransomware and extortion
2021 Microsoft Digital Defense Report (see pages 10-19)
Ransomware: A pervasive and ongoing threat threat analytics report in the Microsoft 365 Defender portal
Microsoft’s DART ransomware approach and best practices
Microsoft 365:
Deploy ransomware protection for your Microsoft 365 tenant
Maximize Ransomware Resiliency with Azure and Microsoft 365
Recover from a ransomware attack
Malware and ransomware protection
Protect your Windows 10 PC from ransomware
Handling ransomware in SharePoint Online
Threat analytics reports for ransomware in the Microsoft 365 Defender portal
Microsoft 365 Defender:
Find ransomware with advanced hunting
Microsoft Azure:
Azure Defenses for Ransomware Attack
Maximize Ransomware Resiliency with Azure and Microsoft 365
Backup and restore plan to protect against ransomware
Help protect from ransomware with Microsoft Azure Backup (26 minute video)
Recovering from systemic identity compromise
Advanced multistage attack detection in Microsoft Sentinel
Fusion Detection for Ransomware in Microsoft Sentinel
Microsoft Defender for Cloud Apps:
Create anomaly detection policies in Defender for Cloud Apps
Microsoft Security team blog posts:
3 steps to prevent and recover from ransomware (September 2021)
A guide to combatting human-operated ransomware: Part 1 (September 2021)
Key steps on how Microsoft's Detection and Response Team (DART) conducts ransomware incident
investigations.
A guide to combatting human-operated ransomware: Part 2 (September 2021)
Recommendations and best practices.
Becoming resilient by understanding cybersecurity risks: Part 4—navigating current threats (May 2021)
See the Ransomware section.
Human-operated ransomware attacks: A preventable disaster (March 2020)
Includes attack chain analyses of actual attacks.
Ransomware response—to pay or not to pay? (December 2019)
Norsk Hydro responds to ransomware attack with transparency (December 2019)
 Phase 3: Make it hard to get in
12/13/2021 • 9 minutes to read • Edit Online

In this phase, you make the attackers work a lot harder to get into your on-premises or cloud infrastructures by
incrementally removing the risks at the points of entry.

IMPORTANT
While many of these will be familiar and/or easy to quickly accomplish, it’s critically important that your work on Phase
3 should not slow down your progress on phases 1 and 2 !

See these sections:

Remote access
Email and collaboration
Endpoints
Accounts

Remote access
Gaining access to your organization's intranet through a remote access connection is an attack vector for
ransomware attackers. Once an on-premises user account is compromised, an attacker is free to roam on an
intranet to gather intelligence, elevate privileges, and install ransomware code. The recent Colonial Pipeline
cyberattack is an example.
Program and project member accountabilities
This table describes the overall protection of your remote access solution from ransomware in terms of a
sponsorship/program management/project management hierarchy to determine and drive results.

L EA D IM P L EM EN TO R A C C O UN TA B IL IT Y

CISO or CIO Executive sponsorship

Program lead on the Central IT Drive results and cross-team

Infrastructure/Network Team collaboration

IT and Security Architects Prioritize component integration into

architectures

Central IT Identity Team Configure Azure AD and conditional

access policies

Central IT Operations Implement changes to environment

Workload Owners Assist with RBAC permissions for app

publishing

Security Policy and Standards Update standards and policy

documents
 L EA D IM P L EM EN TO R A C C O UN TA B IL IT Y

Security Compliance Management Monitor to ensure compliance

User Education Team Update any guidance on workflow

changes and perform education and
change management

Implementation checklist
Apply these best practices to protect your remote access infrastructure from ransomware attackers.

DO N E TA SK DESC RIP T IO N

Maintain software and appliance Attackers use well-known

updates. Avoid missing or neglecting vulnerabilities that have not yet been
manufacturer protections (security patched as attack vectors.
updates, supported status).

Configure Azure AD for existing Zero Trust provides multiple levels of

remote access by including enforcing securing access to your organization.
Zero Trust user and device validation
with Conditional Access.

Configure security for existing third- Take advantage of the built-in security
party VPN solutions (Cisco of your remote access solution.
AnyConnect, Palo Alto Networks
GlobalProtect & Captive Portal,
Fortinet FortiGate SSL VPN, Citrix
NetScaler, Zscaler Private Access (ZPA),
and more).

Deploy Azure Point-to-Site (P2S) VPN Take advantage of integration with

to provide remote access. Azure AD and your existing Azure
subscriptions.

Publish on-premises web apps with Apps published with Azure AD

Azure AD Application Proxy. Application Proxy do not need a
remote access connection.

Secure access to Azure resources with Securely and seamlessly connect to

Azure Bastion. your Azure virtual machines over SSL.

Audit and monitor to find and fix Reduces risk from ransomware
deviations from baseline and potential activities that probe baseline security
attacks (see Detection and Response). features and settings.

Email and collaboration

Implement best practices for email and collaboration solutions to make it more difficult for attackers to abuse
them, while allowing your workers to easily and safely access external content.
Attackers frequently enter the environment by transferring malicious content in with authorized collaboration
tools such as email and file sharing and convincing users to run it. Microsoft has invested in enhanced
mitigations that vastly increase protection against these attack vectors.
Program and project member accountabilities
This table describes the overall protection of your email and collaboration solutions from ransomware in terms
of a sponsorship/program management/project management hierarchy to determine and drive results.

L EA D IM P L EM EN TO R A C C O UN TA B IL IT Y

CISO, CIO, or Identity Director Executive sponsorship

Program lead from the Security Drive results and cross-team

Architecture team collaboration

IT Architects Prioritize component integration into

architectures

Cloud Productivity or End User Team Enable Defender for Office 365, ASR,
and AMSI

Security Architecture / Infrastructure + Configuration assistance

Endpoint

User Education Team Update guidance on workflow changes

Security Policy and Standards Update standards and policy

documents

Security Compliance Management Monitor to ensure compliance

Implementation checklist
Apply these best practices to protect your email and collaboration solutions from ransomware attackers.

DO N E TA SK DESC RIP T IO N

Enable AMSI for Office VBA. Detect Office macro attacks with
endpoint tools like Defender for
Endpoint.

Implement Advanced Email security Email is a common entry point for

using Defender for Office 365 or a attackers.
similar solution.
 DO N E TA SK DESC RIP T IO N

Enable attack surface reduction (ASR) ASR provides additional layers of

rules to block common attack protect specifically targeted at
techniques including: mitigating common attack methods.

- Endpoint abuse such as credential

theft, ransomware activity, and
suspicious use of PsExec and WMI.

- Weaponized Office document activity

such as advanced macro activity,
executable content, process creation,
and process injection initiated by Office
applications.

Note: Deploy these rules in audit

mode first, then assess any negative
impact, and then deploy them in block
mode.

Audit and monitor to find and fix Reduces risk from ransomware
deviations from baseline and potential activities that probe baseline security
attacks (see Detection and Response). features and settings.

Endpoints
Implement relevant security features and rigorously follow software maintenance best practices for computers
and applications, prioritizing applications and server/client operating systems directly exposed to Internet traffic
and content.
Internet-exposed endpoints are a common entry vector that provides attackers access to the organization's
assets. Prioritize blocking common OS and application with preventive controls to slow or stop them from
executing the next stages.
Program and project member accountabilities
This table describes the overall protection of your endpoints from ransomware in terms of a
sponsorship/program management/project management hierarchy to determine and drive results.

L EA D IM P L EM EN TO R A C C O UN TA B IL IT Y

Business leadership accountable for Executive sponsorship (maintenance)

business impact of both downtime and
attack damage

Central IT Operations or CIO Executive sponsorship (others)

Program lead from the Central IT Drive results and cross-team

Infrastructure Team collaboration

IT and Security Architects Prioritize component integration into

architectures

Central IT Operations Implement changes to environment

 L EA D IM P L EM EN TO R A C C O UN TA B IL IT Y

Cloud Productivity or End User Team Enable attack surface reduction

Workload/App Owners Identify maintenance windows for

changes

Security Policy and Standards Update standards and policy

documents

Security Compliance Management Monitor to ensure compliance

Implementation checklist
Apply these best practices to all Windows, Linux, MacOS, Android, iOS, and other endpoints.

DO N E TA SK DESC RIP T IO N

Block known threats with attack Don't let lack of use of these built-in
surface reduction rules, tamper security features be the reason an
protection, and block at first site. attacker entered your organization.

Apply Security Baselines to harden Protect your organization with the

internet-facing Windows servers and minimum level of security and build
clients and Office applications. from there.

Maintain your software so that it is: Attackers are counting on you missing
or neglecting manufacturer updates
- Updated: Rapidly deploy critical and upgrades.
security updates for operating
systems, browsers, & email clients

- Supported: Upgrade operating

systems and software for versions
supported by your vendors.

Isolate, disable, or retire insecure Attackers use known vulnerabilities of

systems and protocols, including legacy devices, systems, and protocols
unsupported operating systems and as entry points into your organization.
legacy protocols.

Block unexpected traffic with host- Some malware attacks reply on

based firewall and network defenses. unsolicited inbound traffic to hosts as
a way of making a connection for an
attack.

Audit and monitor to find and fix Reduces risk from ransomware
deviations from baseline and potential activities that probe baseline security
attacks (see Detection and Response). features and settings.

Accounts
Just as antique skeleton keys won’t protect a house against a modern-day burglar, passwords cannot protect
accounts against common attacks we see today. While multi-factor authentication (MFA) was once a
burdensome extra step, passwordless authentication improves the sign-in experience using biometric
approaches that don’t require your users to remember or type a password. Additionally, a Zero Trust
infrastructure stores information about trusted devices, which reduce prompting for annoying out-of-band MFA
actions.
Starting with high-privilege administrator accounts, rigorously follow these best practices for account security
including using passwordless or MFA.
Program and project member accountabilities
This table describes the overall protection of your accounts from ransomware in terms of a
sponsorship/program management/project management hierarchy to determine and drive results.

L EA D IM P L EM EN TO R A C C O UN TA B IL IT Y

CISO, CIO, or Identity Director Executive sponsorship

Program lead from Identity and Key Drive results and cross-team
Management or Security Architecture collaboration
teams

IT and Security Architects Prioritize component integration into

architectures

Identity and Key Management or Implement configuration changes

Central IT Operations

Security Policy and Standards Update standards and policy

documents

Security Compliance Management Monitor to ensure compliance

User Education Team Update password or sign-in guidance

and perform education and change
management

Implementation checklist
Apply these best practices to protect your accounts from ransomware attackers.

DO N E TA SK DESC RIP T IO N

Enforce strong MFA or passwordless Make it harder for an attacker to

sign-in for all users. Start with perform a credential compromise.
administrator and priority accounts
using one or more of:

- Passwordless authentication with

Windows Hello or the Microsoft
Authenticator app.

- Azure Multi-Factor Authentication.

- Third-party MFA solution.

 DO N E TA SK DESC RIP T IO N

Increase password security: Ensure that attackers can't determine

common passwords or passwords
- For Azure AD accounts, use Azure based on your organization name.
AD Password Protection to detect and
block known weak passwords and
additional weak terms that are specific
to your organization.

- For on-premises Active Directory

Domain Services (AD DS) accounts,
Extend Azure AD Password Protection
to AD DS accounts.

Audit and monitor to find and fix Reduces risk from ransomware
deviations from baseline and potential activities that probe baseline security
attacks (see Detection and Response). features and settings.

Implementation results and timelines

Try to achieve these results within 30 days:
100 % of employees are actively using MFA
100 % deployment of higher password security

Additional ransomware resources

Key information from Microsoft:
The growing threat of ransomware, Microsoft On the Issues blog post on July 20, 2021
Human-operated ransomware
Rapidly protect against ransomware and extortion
2021 Microsoft Digital Defense Report (see pages 10-19)
Ransomware: A pervasive and ongoing threat threat analytics report in the Microsoft 365 Defender portal
Microsoft’s DART ransomware approach and best practices
Microsoft 365:
Deploy ransomware protection for your Microsoft 365 tenant
Maximize Ransomware Resiliency with Azure and Microsoft 365
Recover from a ransomware attack
Malware and ransomware protection
Protect your Windows 10 PC from ransomware
Handling ransomware in SharePoint Online
Threat analytics reports for ransomware in the Microsoft 365 Defender portal
Microsoft 365 Defender:
Find ransomware with advanced hunting
Microsoft Azure:
Azure Defenses for Ransomware Attack
Maximize Ransomware Resiliency with Azure and Microsoft 365
Backup and restore plan to protect against ransomware
 Help protect from ransomware with Microsoft Azure Backup (26 minute video)
Recovering from systemic identity compromise
Advanced multistage attack detection in Microsoft Sentinel
Fusion Detection for Ransomware in Microsoft Sentinel
Microsoft Defender for Cloud Apps:
Create anomaly detection policies in Defender for Cloud Apps
Microsoft Security team blog posts:
3 steps to prevent and recover from ransomware (September 2021)
A guide to combatting human-operated ransomware: Part 1 (September 2021)
Key steps on how Microsoft's Detection and Response Team (DART) conducts ransomware incident
investigations.
A guide to combatting human-operated ransomware: Part 2 (September 2021)
Recommendations and best practices.
Becoming resilient by understanding cybersecurity risks: Part 4—navigating current threats (May 2021)
See the Ransomware section.
Human-operated ransomware attacks: A preventable disaster (March 2020)
Includes attack chain analyses of actual attacks.
Ransomware response—to pay or not to pay? (December 2019)
Norsk Hydro responds to ransomware attack with transparency (December 2019)
 Backup and restore plan to protect against
ransomware
12/13/2021 • 18 minutes to read • Edit Online

Ransomware attacks deliberately encrypt or erase data and systems to force your organization to pay money to
attackers. These attacks target your data, your backups, and also key documentation required for you to recover
without paying the attackers (as a means to increase the chances your organization will pay).
This article addresses what to do before an attack to protect your critical business systems and during an attack
to ensure a rapid recovery of business operations.

NOTE
Preparing for ransomware also improves resilience to natural disasters and rapid attacks like WannaCry & (Not)Petya.

What is ransomware?
Ransomware is a type of extortion attack that encrypts files and folders, preventing access to important data
and systems. Attackers use ransomware to extort money from victims by demanding money, usually in the form
of cryptocurrencies, in exchange for a decryption key or in exchange for not releasing sensitive data to the dark
web or the public internet.
While early ransomware mostly used malware that spread with phishing or between devices, human-operated
ransomware has emerged where a gang of active attackers, driven by human attack operators, target all
systems in an organization (rather than a single device or set of devices). An attack can:
Encrypt your data
Exfiltrate your data
Corrupt your backups
The ransomware leverages the attackers’ knowledge of common system and security misconfigurations and
vulnerabilities to infiltrate the organization, navigate the enterprise network, and adapt to the environment and
its weaknesses as they go.
Ransomware can be staged to exfiltrate your data first, over several weeks or months, before the ransomware
actually executes on a specific date.
Ransomware can also slowly encrypt your data while keeping your key on the system. With your key still
available, your data is usable to you and the ransomware goes unnoticed. Your backups, though, are of the
encrypted data. Once all of your data is encrypted and recent backups are also of encrypted data, your key is
removed so you can no longer read your data.
The real damage is often done when the attack exfiltrates files while leaving backdoors in the network for future
malicious activity—and these risks persist whether or not the ransom is paid. These attacks can be catastrophic
to business operations and difficult to clean up, requiring complete adversary eviction to protect against future
attacks. Unlike early forms of ransomware that only required malware remediation, human-operated
ransomware can continue to threaten your business operations after the initial encounter.
Impact of an attack
The impact of a ransomware attack on any organization is difficult to quantify accurately. Depending on the
scope of the attack, the impact could include:
Loss of data access
Business operation disruption
Financial loss
Intellectual property theft
Compromised customer trust or tarnished reputation
Legal expenses

How can you protect yourself?

The best way to prevent falling victim to ransomware is to implement preventive measures and have tools that
protect your organization from every step that attackers take to infiltrate your systems.
You can reduce your on-premises exposure by moving your organization to a cloud service. Microsoft has
invested in native security capabilities that make Microsoft Azure resilient against ransomware attacks and helps
organizations defeat ransomware attack techniques. For a comprehensive view of ransomware and extortion
and how to protect your organization, use the information in the Human-Operated Ransomware Mitigation
Project Plan PowerPoint presentation.
You should assume that at some point in time you will fall victim to a ransomware attack. One of the most
important steps you can take to protect your data and avoid paying a ransom is to have a reliable backup and
restore plan for your business-critical information. Since ransomware attackers have invested heavily into
neutralizing backup applications and operating system features like volume shadow copy, it is critical to have
backups that are inaccessible to a malicious attacker.
Azure Backup
Azure Backup provides security to your backup environment, both when your data is in transit and at rest. With
Azure Backup, you can back up:
On-premises files, folders, and system state
Entire Windows/Linux VMs
Azure Managed Disks
Azure file shares to a storage account
SQL Server databases running on Azure VMs
The backup data is stored in Azure storage and the guest or attacker has no direct access to backup storage or
its contents. With virtual machine backup, the backup snapshot creation and storage is done by Azure fabric
where the guest or attacker has no involvement other than quiescing the workload for application consistent
backups. With SQL and SAP HANA, the backup extension gets temporary access to write to specific blobs. In this
way, even in a compromised environment, existing backups can't be tampered with or deleted by the attacker.
Azure Backup provides built-in monitoring and alerting capabilities to view and configure actions for events
related to Azure Backup. Backup Reports serve as a one-stop destination for tracking usage, auditing of backups
and restores, and identifying key trends at different levels of granularity. Using Azure Backup's monitoring and
reporting tools can alert you to any unauthorized, suspicious, or malicious activity as soon as they occur.
Checks have been added to make sure only valid users can perform various operations. These include adding an
extra layer of authentication. As part of adding an extra layer of authentication for critical operations, you're
prompted to enter a security PIN before modifying online backups.
Learn more about the security features built into Azure Backup.
Validate backups
Validate that your backup is good as your backup is created and before you restore. We recommend that you
use a Recovery Services vault, which is a storage entity in Azure that houses data. The data is typically copies of
data, or configuration information for virtual machines (VMs), workloads, servers, or workstations. You can use
Recovery Services vaults to hold backup data for various Azure services such as IaaS VMs (Linux or Windows)
and Azure SQL databases as well as on-premises assets. Recovery Services vaults make it easy to organize your
backup data and provide features such as:
Enhanced capabilities to ensure you can secure your backups, and safely recover data, even if production and
backup servers are compromised. Learn more.
Monitoring for your hybrid IT environment (Azure IaaS VMs and on-premises assets) from a central portal.
Learn more.
Compatibility with Azure role-based access control (Azure RBAC), which restricts backup and restore access
to a defined set of user roles. Azure RBAC provides various built-in roles, and Azure Backup has three built-in
roles to manage recovery points. Learn more.
Soft delete protection, even if a malicious actor deletes a backup (or backup data is accidentally deleted).
Backup data is retained for 14 additional days, allowing the recovery of a backup item with no data loss.
Learn more.
Cross Region Restore which allows you to restore Azure VMs in a secondary region, which is an Azure paired
region. You can restore the replicated data in the secondary region any time. This enables you to restore the
secondary region data for audit-compliance, and during outage scenarios, without waiting for Azure to
declare a disaster (unlike the GRS settings of the vault). Learn more.

NOTE
There are two types of vaults in Azure Backup. In addition to the Recovery Services vaults, there are also Backup vaults
that house data for newer workloads supported by Azure Backup.

What to do before an attack

As mentioned earlier, you should assume that at some point in time you will fall victim to a ransomware attack.
Identifying your business-critical systems and applying best practices before an attack will get you back up and
running as quickly as possible.
Determine what is most important to you
Ransomware can attack while you are planning for an attack so your first priority should be to identify the
business-critical systems that are most important to you and begin performing regular backups on those
systems.
In our experience, the five most important applications to customers fall into the following categories in this
priority order:
Identity systems – required for users to access any systems (including all others described below) such as
Active Directory, Azure AD Connect, AD domain controllers
Human life – any system that supports human life or could put it at risk such as medical or life support
systems, safety systems (ambulance, dispatch systems, traffic light control), large machinery,
chemical/biological systems, production of food or personal products, and others
Financial systems – systems that process monetary transactions and keep the business operating, such as
payment systems and related databases, financial system for quarterly reporting
Product or service enablement – any systems that are required to provide the business services or
produce/deliver physical products that your customers pay you for, factory control systems, product
delivery/dispatch systems, and similar
Security (minimum) – You should also prioritize the security systems required to monitor for attacks and
provide minimum security services. This should be focused on ensuring that the current attacks (or easy
 opportunistic ones) are not immediately able to gain (or regain) access to your restored systems
Your prioritized back up list also becomes your prioritized restore list. Once you’ve identified your critical
systems and are performing regular backups, then take steps to reduce your exposure level.
Steps to take before an attack
Apply these best practices before an attack.

TA SK DETA IL

Identify the important systems that you need to bring back To get back up and running as quickly as possible after an
online first (using top five categories above) and immediately attack, determine today what is most important to you.
begin performing regular backups of those systems.

Migrate your organization to the cloud. Reduce your on-premises exposure by moving data to cloud
services with automatic backup and self-service rollback.
Consider purchasing a Microsoft Unified Support plan or Microsoft Azure has a robust set of tools to help you backup
working with a Microsoft partner to help support your move your business-critical systems and restore your backups
to the cloud. faster.

Microsoft Unified Support is a cloud services support model

that is there to help you whenever you need it. Unified
Support:

Provides a designated team that is available 24x7 with as-

needed problem resolution and critical incident escalation

Helps you monitor the health of your IT environment and

works proactively to make sure problems are prevented
before they happen

Move user data to cloud solutions like OneDrive and User data in the Microsoft cloud can be protected by built-in
SharePoint to take advantage of versioning and recycle bin security and data management features.
capabilities.
It's good to teach users how to restore their own files but
Educate users on how to recover their files by themselves to you need to be careful that your users do not restore the
reduce delays and cost of recovery. For example, if a user’s malware used to carry out the attack. You need to:
OneDrive files were infected by malware, they can restore
their entire OneDrive to a previous time. Ensure your users don't restore their files until you are
confident that the attacker has been evicted
Consider a defense strategy, such as Microsoft 365
Defender, before allowing users to restore their own files. Have a mitigation in place in case a user does restore some
of the malware

Microsoft 365 Defender uses AI-powered automatic actions

and playbooks to remediate impacted assets back to a
secure state. Microsoft 365 Defender leverages automatic
remediation capabilities of the suite products to ensure all
impacted assets related to an incident are automatically
remediated where possible.

Implement Azure Security Benchmark. Azure Security Benchmark is Azure’s own security control
framework based on industry-based security control
frameworks such as NIST SP800-53, CIS Controls v7.1. It
provides organizations guidance on how to configure Azure
and Azure services and implement the security controls. See
Backup and Recovery.
TA SK DETA IL

Regularly exercise your business continuity/disaster recovery Ensures rapid recovery of business operations by treating a
(BC/DR) plan. ransomware or extortion attack with the same importance
as a natural disaster.
Simulate incident response scenarios. Exercises you perform
in preparing for an attack should be planned and conducted Conduct practice exercise(s) to validate cross-team processes
around your prioritized backup and restore lists. and technical procedures, including out of band employee
and customer communications (assume all email and chat is
Regularly test ‘Recover from Zero’ scenario to ensure your down).
BC/DR can rapidly bring critical business operations online
from zero functionality (all systems down).

Consider creating a risk register to identify potential risks A risk register can help you prioritize risks based on the
and address how you will mediate through preventative likelihood of that risk occurring and the severity to your
controls and actions. Add ransomware to risk register as business should that risk occur.
high likelihood and high impact scenario.
Track mitigation status via Enterprise Risk Management
(ERM) assessment cycle.

Backup all critical business systems automatically on a Allows you to recover data up to the last backup.
regular schedule (including backup of critical dependencies
like Active Directory).

Validate that your backup is good as your backup is created.

Protect (or print) supporting documents and systems Attackers deliberately target these resources because it
required for recovery such as restoration procedure impacts your ability to recover.
documents, CMDB, network diagrams, and SolarWinds
instances.

Ensure you have well-documented procedures for engaging Third-party contacts may be useful if the given ransomware
any third-party support, particularly support from threat variant has known weaknesses or decryption tools are
intelligence providers, antimalware solution providers, and available.
from the malware analysis provider. Protect (or print) these
procedures.

Ensure backup and recovery strategy includes: Backups are essential for resilience after an organization has
been breached. Apply the 3-2-1 rule for maximum
Ability to back up data to a specific point in time. protection and availability: 3 copies (original + 2 backups), 2
storage types, and 1 offsite or cold copy.
Multiple copies of backups are stored in isolated, offline (air-
gapped) locations.

Recovery time objectives that establish how quickly backed

up information can be retrieved and put into production
environment.

Rapid restore of back up to a production

environment/sandbox.
TA SK DETA IL

Protect backups against deliberate erasure and encryption: Backups that are accessible by attackers can be rendered
unusable for business recovery.
Store backups in offline or off-site storage and/or immutable
storage. Offline storage ensures robust transfer of backup data
without using any network bandwidth. Azure Backup
Require out of band steps (such as MFA or a security PIN) supports offline backup, which transfers initial backup data
before permitting an online backup to be modified or erased. offline, without the use of network bandwidth. It provides a
mechanism to copy backup data onto physical storage
Create private endpoints within your Azure Virtual Network devices. The devices are then shipped to a nearby Azure
to securely back up and restore data from your Recovery datacenter and uploaded onto a Recovery Services vault.
Services vault.
Online immutable storage (such as Azure Blob) enables you
to store business-critical data objects in a WORM (Write
Once, Read Many) state. This state makes the data non-
erasable and non-modifiable for a user-specified interval.

Multi-factor authentication (MFA) should be mandatory for

all admin accounts and is strongly recommended for all
users. The preferred method is to use an authenticator app
rather than SMS or voice where possible. When you set up
Azure Backup you can configure your recovery services to
enable MFA using a security PIN generated in the Azure
portal. This ensures that a security pin is generated to
perform critical operations such as updating or removing a
recovery point.

Designate protected folders. Makes it more difficult for unauthorized applications to

modify the data in these folders.

Review your permissions: Reduces risk from broad access-enabling ransomware

activities.
Discover broad write/delete permissions on file shares,
SharePoint, and other solutions. Broad is defined as many
users having write/delete permissions for business-critical
data.

Reduce broad permissions while meeting business

collaboration requirements.

Audit and monitor to ensure broad permissions don’t

reappear.

Protect against a phishing attempt: The most common method used by attackers to infiltrate an
organization is phishing attempts via email. Exchange Online
Conduct security awareness training regularly to help users Protection (EOP) is the cloud-based filtering service that
identify a phishing attempt and avoid clicking on something protects your organization against spam, malware, and other
that can create an initial entry point for a compromise. email threats. EOP is included in all Microsoft 365
organizations with Exchange Online mailboxes.
Apply security filtering controls to email to detect and
minimize the likelihood of a successful phishing attempt. An example of a security filtering control for email is Safe
Links. Safe Links is a feature in Defender for Office 365 that
provides URL scanning and rewriting of inbound email
messages in mail flow, and time-of-click verification of URLs
and links in email messages and other locations. Safe Links
scanning occurs in addition to the regular anti-spam and
anti-malware protection in inbound email messages in EOP.
Safe Links scanning can help protect your organization from
malicious links that are used in phishing and other attacks.

Learn more about anti-phishing protection.

What to do during an attack
If you are attacked, your prioritized back up list becomes your prioritized restore list. Before you restore, validate
again that your backup is good. You may be able to look for malware inside the backup.
Steps to take during an attack
Apply these best practices during an attack.

TA SK DETA IL

Early in the attack, engage third-party support, particularly These contacts may be useful if the given ransomware
support from threat intelligence providers, antimalware variant has a known weakness or decryption tools are
solution providers and from the malware analysis provider. available.

Microsoft Detection and Response Team (DART) engages

with customers globally to identify risks and provide reactive
incident response and proactive security investigation
services. The DART helps our customers manage their cyber-
risk, especially in today’s dynamic threat environment.

Microsoft also provides Rapid Ransomware Recovery

services. Services are exclusively delivered by the Microsoft
Global Compromise Recovery Security Practice (CRSP). The
focus of this team during a ransomware attack is to restore
authentication service and limit the impact of ransomware.

DART and CRSP are part of Microsoft’s Industry Solutions

Delivery security service line.

Contact your local or federal law enforcement agencies. If you are in the United States, contact the FBI to report a
ransomware breach using the IC3 Complaint Referral Form.

Take steps to remove malware or ransomware payload from You can use Windows Defender or (for older clients)
your environment and stop the spread. Microsoft Security Essentials.

Run a full, current antivirus scan on all suspected computers An alternative that will also help you remove ransomware or
and devices to detect and remove the payload that's malware is the Malicious Software Removal Tool (MSRT).
associated with the ransomware.

Scan devices that are synchronizing data, or the targets of

mapped network drives.

Restore business-critical systems first. Remember to validate At this point, you don’t need to restore everything. Focus on
again that your backup is good before you restore. the top five business-critical systems from your restore list.

If you have offline backups, you can probably restore the To prevent future attacks, ensure ransomware or malware is
encrypted data after you've removed the ransomware not on your offline backup before restoring.
payload (malware) from your environment.

Identify a safe point-in-time backup image that is known not To prevent future attacks, scan backup for ransomware or
to be infected. malware before restoring.

If you use Recovery Services vault, carefully review the

incident timeline to understand the right point-in-time to
restore a backup.

Use a safety scanner and other tools for full operating Microsoft Safety Scanner is a scan tool designed to find and
system restore as well as data restore scenarios. remove malware from Windows computers. Simply
download it and run a scan to find malware and try to
reverse changes made by identified threats.
 TA SK DETA IL

Ensure that your antivirus or endpoint detection and An EDR solution, such as Microsoft Defender for Endpoint, is
response (EDR) solution is up to date. You also need to have preferred.
up-to-date patches.

After business-critical systems are up and running, restore Telemetry data should help you identify if malware is still on
other systems. your systems.

As systems get restored, start collecting telemetry data so

you can make formative decisions about what you are
restoring.

Post attack or simulation

After a ransomware attack or an incident response simulation, take the following steps to improve your backup
and restore plans as well as your security posture:
1. Identify lessons learned where the process did not work well (and opportunities to simplify, accelerate, or
otherwise improve the process)
2. Perform root cause analysis on the biggest challenges (at enough detail to ensure solutions address the right
problem — considering people, process, and technology)
3. Investigate and remediate the original breach (engage the Microsoft Detection and Response Team (DART) to
help)
4. Update your backup and restore strategy based on lessons learned and opportunities — prioritizing based
on highest impact and quickest implementation steps first

Next steps
In this article, you learned how to improve your backup and restore plan to protect against ransomware. For
best practices on deploying ransomware protection, see Rapidly protect against ransomware and extortion.
Key industry information:
2021 Microsoft Digital Defense Report (see pages 10-19)
Microsoft Azure:
Help protect from ransomware with Microsoft Azure Backup (26 minute video)
Recovering from systemic identity compromise
Advanced multistage attack detection in Microsoft Sentinel
Microsoft 365:
Recover from a ransomware attack
Malware and ransomware protection
Protect your Windows 10 PC from ransomware
Handling ransomware in SharePoint Online
Microsoft 365 Defender:
Find ransomware with advanced hunting
Microsoft Security team blog posts:
Becoming resilient by understanding cybersecurity risks: Part 4, navigating current threats (May 2021). See
the Ransomware section
Human-operated ransomware attacks: A preventable disaster (March 2020). Includes attack chain analysis of
actual human-operated ransomware attacks
Ransomware response — to pay or not to pay? (December 2019)
Norsk Hydro responds to ransomware attack with transparency (December 2019)
 Microsoft Security Best Practices module:
Information protection and storage
12/13/2021 • 2 minutes to read • Edit Online

Intellectual property that is valuable to the organization (or its customers/constituents) requires security
protection appropriate to its value.
See the Storage, data, and encryption and Capabilities topics for more information.
The following videos provide guidance on information protection and storage. You can also download the
PowerPoint slides associated with these videos.
For more information about information protection capabilities across Microsoft 365 and SQL databases, see
CISO Workshop Module 5: Information Protection and Information protection and storage capabilities.

Part 1: Introduction (13:39)

Part 2: Storage and Encryption Best Practices (03:30)

Next steps
For additional security guidance from Microsoft, see Microsoft security documentation.
 Storage, data, and encryption
12/13/2021 • 3 minutes to read • Edit Online

Protecting data at rest is required to maintain confidentiality, integrity, and availability assurances across all
workloads. Storage in a cloud service like Azure is architected and implemented quite differently than on
premises solutions to enable massive scaling, modern access through REST APIs, and isolation between tenants.
Granting access to Azure storage is possible through Azure Active Directory (Azure AD) as well as key based
authentication mechanisms (Symmetric Shared Key Authentication, or Shared Access Signature (SAS))
Storage in Azure includes a number of native security design attributes
All data is encrypted by the service
Data in the storage system cannot be read by a tenant if it has not been written by that tenant (to mitigate
the risk of cross tenant data leakage)
Data will remain only in the region you choose
The system maintains three synchronous copies of data in the region you choose.
Detailed activity logging is available on an opt-in basis.
Additional security features can be configured such as a storage firewall to provide an additional layer of access
control as well as storage threat protection to detect anomalous access and activities.
Encryption is a powerful tool for security, but it's critical to understand its limits in protecting data. Much like a
safe, encryption restricts access to only those with possession of a small item (a mathematical key). While it's
easier to protect possession of keys than larger datasets, it is imperative that you provide the appropriate
protections for the keys. Protecting cryptographic keys is not a natural intuitive human process (especially
because electronic data like keys can be copied perfectly without a forensic evidence trail), so it is often
overlooked or implemented poorly.
While encryption is available in many layers in Azure (and often on by default), we have identified the layers that
are most important to implement (high potential for data to move to another storage medium) and are easiest
to implement (near zero overhead).

Use Identity based storage access controls

Cloud service providers make multiple methods of access control over storage resources available. Examples
include shared keys, shared signatures, anonymous access, and identity provider-based methods.
Identify provider methods of authentication and authorization are the least liable to compromise and enable
more fine-grained role-based access controls over storage resources.
We recommend that you use an identity-based option for storage access control.
An example of this is Azure Active Directory Authentication to Azure blob and queue services.

Encrypt virtual disk files

Virtual machines use virtual disk files as virtual storage volumes and exist in a cloud service provider’s blob
storage system. These files can be moved from on-premises to cloud systems, from cloud systems to on-
premises, or between cloud systems. Due to the mobility of these files, you need to make sure the files and their
contents are not accessible to unauthorized users.
Authentication-based access controls should be in place to prevent potential attackers from downloading the
files to their own systems. In the event of a flaw in the authentication and authorization system or its
configuration, you want to have a backup mechanism to secure the virtual disk files.
You can encrypt the virtual disk files to help prevent attackers from gaining access to the contents of the disk
files in the event that an attacker is able to download the files. When attackers attempt to mount an encrypted
disk file, they will not be able to because of the encryption.
We recommend that you enable virtual disk encryption.
An example of virtual disk encryption is Azure Disk Encryption.

Enable platform encryption services

All public cloud service providers enable encryption that is done automatically using provider-managed keys on
their platform. In many cases, this is done for the customer and no user interaction is required. In other cases,
the provider makes this an option that the customer can choose to use or not to use.
There is almost no overhead in enabling this type of encryption as it’s managed by the cloud service provider.
We recommend that for each service that supports service provider encryption that you enable that option.
An example of service-specific service provider encryption is Azure Storage Service encryption.

Next steps
For additional security guidance from Microsoft, see Microsoft security documentation.
 Information protection and storage capabilities
12/13/2021 • 3 minutes to read • Edit Online

This article lists the capabilities that can help with information protection and storage.

Capabilities that work with Microsoft 365

Microsoft 365 and Office 365 include capabilities that can be applied to specific types of data to protect
information across Microsoft 365 tools, including OneDrive, SharePoint Online, and mail. Some capabilities, like
sensitive information types, can be used with Microsoft Defender for Cloud Apps to protect information across
other SaaS apps in your environment.

C A PA B IL IT Y M O RE IN F O RM AT IO N

Sensitivity labels With sensitivity labels you can classify

and help protect your sensitive
content. Protection options include
labels, watermarks, and encryption.
Sensitivity labels use Azure Information
Protection. If you are using Azure
Information Protection labels, for now
we recommend that you avoid
creating new labels in other admin
centers until after you've completed
your migration. See Azure Information
Protection migration.
Retention labels are different than
sensitivity labels. Retention labels help
you retain or delete content based on
policies that you define. These help
organizations comply with industry
regulations and internal policies.

Data loss prevention (DLP) With DLP policies, you can identify,
monitor, and automatically protect
sensitive information across Office 365.
Data loss prevention policies can use
sensitivity labels and sensitive
information types to identify sensitive
information.

Sensitive information types Microsoft 365 includes many sensitive

information types that are ready for
you to use in DLP policies and for
automatic classification with sensitivity
and retention labels. Sensitive
information types can also be used
with the Azure Information Protection
scanner to classify and protect files on
premises. Sensitive information types
define how the automated process
recognizes specific information types
such as health service numbers and
credit card numbers.
 C A PA B IL IT Y M O RE IN F O RM AT IO N

Office 365 Message Encryption (OME) With Office 365 Message Encryption,
your organization can send and receive
encrypted email messages between
people inside and outside your
organization. Office 365 Message
Encryption works with Outlook.com,
Yahoo!, Gmail, and other email
services. Email message encryption
helps ensure that only intended
recipients can view message content.

Azure Information Protection Azure Information Protection

(sometimes referred to as AIP) helps
an organization to classify, label, and
optionally, protect documents and
emails. Administrators can
automatically apply labels by defining
rules and conditions. Users can
manually apply labels to files and mail.
You can also give users
recommendations about when to
apply labels.
If you're using sensitivity labels or
Office Message Encryption, you're
already using classification and
protection capabilities. If you haven't
yet migrated Azure Information
Protection labels to Office 365,
continue to manage these in Azure
Information Protection.
You can run the Azure Information
Protection scanner on premises to
classify and protect files on Windows
Server, network shares, and SharePoint
Server sites and libraries. This can be a
first step toward identifying data to
migrate to Office 365.

Azure Information Protection with Some organizations have a business

customer managed encryption key need or compliance requirement to
retain control of an encryption key.
This is not common. Azure Information
Protection allows organizations to
bring your own key (BYOK) to the
service. For more information, see
Bring your own key (BYOK) for Azure
Information Protection. Another more
complex option is offered for
customers who have a requirement to
retain an encryption key on premises,
referred to as hold your own key
(HYOK). For more information, see
Hold your own key (HYOK) for Azure
Information Protection.

Azure storage and encryption

 C A PA B IL IT Y DESC RIP T IO N M O RE IN F O RM AT IO N

Azure Storage Azure Storage includes Azure Blobs Azure Storage documentation
(objects), Azure Data Lake Storage
Gen2, Azure Files, Azure Queues, and
Azure Tables.

Azure encryption Azure encryption options include Azure encryption overview

encryption at rest, encryption in flight,
and key management with Azure Key
Vault

Azure SQL Database

C A PA B IL IT Y DESC RIP T IO N M O RE IN F O RM AT IO N

Azure SQL Database Azure SQL Database is a general- Azure SQL Database documentation
purpose relational database, provided
as a managed service. With it, you can
create a highly available and high-
performance data storage layer for the
applications and solutions in Azure.

Azure SQL Database security Security capabilities for data include An overview of Azure SQL Database
capabilities Always encrypted and Transparent security capabilities
Data Encryption (TDE)

Next steps
For additional security guidance from Microsoft, see Microsoft security documentation.
 Applications and services
12/13/2021 • 16 minutes to read • Edit Online

Applications and the data associated with them ultimately act as the primary store of business value on a cloud
platform. While the platform components like identity and storage are critical elements of the security
environment, applications play an outsize role in risks to the business because:
Business Processes are encapsulated and executed by applications and services need to be available
and provided with high integrity
Business Data is stored and processed by application workloads and requires high assurances of
confidentiality, integrity, and availability.
This section focuses on applications written by your organization or by others on behalf of your organization vs.
SaaS or commercially available applications installed on IaaS VMs.

Modern cloud platforms like Azure can host both legacy and modern generations of applications
Legacy applications are hosted on Infrastructure as a Service (IaaS) virtual machines that typically
include all dependencies including OS, middleware, and other components.
Modern Platform as a Service (PaaS) applications don’t require the application owner to manage and
secure the underlying server operating systems (OSes) and are sometimes fully “Serverless” and built
primarily using functions as a service.
Notes: Popular forms of modern applications are application code hosted on Azure App Services and
containerized applications (though containers can also be hosted on IaaS VMs or on-premises as well).
Hybrid – While hybrid applications can take many forms, the most common is an “IaaS plus” state where
legacy applications are transitioning to a modern architecture with modern services replacing legacy
components or being added a legacy application.
Securing an application requires security assurances for three different component types:
 Application Code – This is the logic that defines the custom application that you write. The security of
this code is the application owners’ responsibility in all generations of application architecture including
any open-source snippets or components included in the code. Securing the code requires identifying
and mitigating risks from the design and implementation of the application as well as assessing supply
chain risk of included components. Note that the evolution of applications into microservices
architectures will break various aspects of application code into smaller services vs. a single monolithic
codebase.
Application Ser vices – These are the various standardized components that the application uses such
as databases, identity providers, event hubs, IoT device management, and so on. For cloud services this is
a shared responsibility:
Cloud Provider - The security of the underlying service is the responsibility of the cloud provider
Application Owner - The application owner is responsible for security implications of the
configuration and operation of the service instance(s) used by the application including any data
stored and processed on the service.
Application Hosting Platform – This is the computing environment where the application actually
executes and runs. In an enterprise with applications hosted on premises, in Azure and in third-party
clouds like Amazon Web Services (AWS), this could take many forms with significant variations on who is
responsible for security:
Legacy Applications typically require a full operating system (and any middleware) hosted on
physical or virtualized hardware. The virtual hardware can be hosted on premises or on
Infrastructure as a Service (IaaS) VMs. This operating system and installed middleware/other
components are operated and secured by the application owner or their infrastructure team(s).
The responsibility for the physical hardware and OS virtualization components (virtualization
hosts, operating systems, and management services) varies:
On premises - The application owner or their organization is responsible for maintenance
and security.
IaaS – The cloud provider is responsible for maintenance and security of the underlying
infrastructure and the application owner’s organization is responsible for the VM
configuration, operating system, and any components installed on it.
Modern Applications are hosted on Platform as a Service (PaaS) environments such as an Azure
application service. In most application service types, the underlying operating system is
abstracted from the application owner and secured by the cloud provider. Application owners are
responsible for the security of the application service configurations that are provided to them.
Containers are an application packaging mechanism in which applications are abstracted from
the environment in which they run. These containerized applications fit into either the legacy or
modern models above depending on whether they are run on a container service by the cloud
provider (Modern Applications) or on a server managed by the organization (on premises or in
IaaS). See the container security section below for more details.

Identify and classify business critical applications

Ensure you have identified and classified the applications in your portfolio that are critical to business functions.
Enterprise organizations typically have a large application portfolio, so prioritizing where to invest time and
effort into manual and resource-intensive tasks like threat modeling can increase the effectiveness of your
security program.
Identify applications that have a high potential impact and/or a high potential exposure to risk.
 High potential impact – Identify application that would a significant impact on the business if
compromised. This could take the form of one or more of:
Business critical data – Applications that process or store information, which would cause
significant negative business or mission impact if an assurance of confidentiality, integrity, or
availability is lost.
Regulated data – Applications that handle monetary instruments and sensitive personal
information regulated by standards. For example, payment card industry (PCI) and Health
Information Portability and Accountability Act (HIPAA).
Business critical availability – Applications whose functionality is critical to organizations
business mission such as production lines generating revenue, devices, or services critical to life
and safety, and other critical functions.
Significant Access – Applications which have access to systems with a high potential impact
through technical means such as
Stored Credentials or keys/certificates that grant access to the data/service
Permissions granted via access control lists or other means
High exposure to attacks – Applications that are easily accessible to attackers such as web applications
on the open internet. Legacy applications can also be higher exposure as attackers and penetration
testers frequently target them because they know these legacy applications often have vulnerabilities that
are difficult to fix.

Adopt the DevOps approach

Organizations should shift from a ‘Waterfall’ development cycle to DevOps lifecycle of continuous integration,
continuous delivery (CI/CD) for applications as fast as is practical. DevOps is the union of people, processes, and
tools that enable continuous delivery of value to end users. The contraction of Dev and Ops refers to combining
the development and operations disciplines into multi-disciplinary teams that work together with shared and
efficient practices and tools.
The DevOps model increases the organization’s ability to rapidly address security concerns without waiting for a
longer planning and testing cycle of a waterfall model.

Follow DevOps security guidance

Organizations should leverage guidance and automation for securing applications on the cloud rather than
starting from zero.
Using resources and lessons learned by external organizations that are early adopters of these models can
accelerate the improvement of an organization’s security posture with less expenditure of effort and resources.
Microsoft has released a toolkit for Secure DevOps on Azure –
https://azsk.azurewebsites.net/
Organization for Web App Security Project (OWASP) has published guidance DevOps Pipeline security
https://www.owasp.org/index.php/OWASP_AppSec_Pipeline#tab=Main

Use Cloud services instead of custom implementations

Developers should use services available from your cloud provider for well-established functions like databases,
encryption, identity directory, and authentication instead of writing custom versions of them.
These services provide better security, reliability, and efficiency because cloud providers operate and secure
them with dedicated teams with deep expertise in those areas. Using these services also frees your developer
resources from reinventing the proverbial wheel so that they can focus development time on your unique
requirements for your business. This practice should be followed to avoid risk during new application
development as well as to reduce risk in existing applications either during planned update cycle or with a
security-focused application update.
Several capabilities that should be prioritized first because of potential security impact:
Identity – User directories and other authentication functions are complex to develop and critically
important to security assurances. Avoid using homegrown authentication solutions and favor mature
capabilities like Azure Active Directory (Azure AD), Azure AD B2B, Azure AD B2C, or third-party solutions
to authenticate and grant permission to users, partners, customers, applications, services, and other
entities.
Data Protection – Developers should use established capabilities from cloud providers such as native
encryption in cloud services to encrypt and protect data. The security world is littered with examples of
failed attempts to protect data or passwords that didn’t stand up to real world attacks. If direct use of
cryptography is required, developers should call well-established cryptographic algorithms and not
attempt to invent their own.
Key management – Ideally use identity for authentication rather than directly handling keys (see Prefer
Identity Authentication over Keys). For situations where accessing services that require access to keys,
leverage a key management service like Azure Key Vault or AWS Key Management Service to manage
and secure these keys rather than attempting to safely handle keys in application code. You can use
CredScan to discover potentially exposed keys in your application code.
Application Configurations – Inconsistent configurations for applications can create security Risks.
Azure App Configuration provides a service to centrally manage application settings and feature flags,
which helps mitigate this risk.

Use Native Security capabilities in application services

Use native security capabilities built into cloud services instead of adding external security components (for data
encryption, network traffic filtering, threat detection, and other functions).
Native security controls are maintained and supported by the service provider, eliminating or reducing effort
required to integrate external security tooling and update those integrations over time. Cloud services evolve
rapidly, which greatly increases the burden of maintaining an external tool and increases risk of losing security
visibility and protections from these tools if the tool doesn’t keep up with the cloud service.
List of Azure Services
https://azure.microsoft.com/services/
Native security capabilities of each service
/azure/security/common-security-attributes

Prefer Identity Authentication over Keys

Always authenticate with identity services rather than cryptographic keys when available.
Managing keys securely with application code is difficult and regularly leads to mistakes like accidentally
publishing sensitive access keys to code repositories like GitHub. Identity systems offer secure and usable
experience for access control with built-in sophisticated mechanisms for key rotation, monitoring for anomalies,
and more. Most organizations also have skilled teams dedicated to managing identity systems and few (if any)
people actively managing key security systems.
For services that offer the Azure AD authentication like Azure Storage, Azure App Service, Azure Backup, use it
for authentication and authorization. To further simplify using identities for developers, you can also take
advantage of managed identities to assign identities to resources like VMs and App Services so that developers
don’t have to manage identities within the application.

Bottom-up approach to reduce security bug volume and impact

Reduce the count and potential severity of security bugs in your application by implementing security practices
and tools during the development lifecycle.
Security bugs can result in an application disclosing confidential data, allowing criminals to alter data/records, or
the data/application becoming unavailable for use by customers and employees. Applications will always have
some logic errors that can result in security risk, so it is important to discover, evaluate, and correct them to
avoid damage to the organization’s reputation, revenue, or margins. It is easier and cheaper to resolve these
earlier in the development lifecycle than it is to correct them after application has completed testing, is in
production use, or has been breached frequently called “shift left” or “push left” principle.
Mitigating application risk is achieved by integrating security practices and tools into the development lifecycle,
often called a secure development lifecycle (SDL or SDLC). Microsoft has published a number of
recommendations in a whitepaper entitled Develop Secure Apps on Azure based on Microsoft’s Security
Development Lifecycle to mitigate common risks with input and output validation, perform fuzz testing, attack
surface reviews, and more.

Top-down approach through threat modeling

Perform threat modeling on your business-critical applications to discover and mitigate potential risks to your
organization.
Threat modeling identifies risks to the application itself as well as risks that application may pose to your
enterprise particularly when evaluating individual applications in a larger system.
Threat modeling can be used at any stage of application development or production, but it is uniquely effective
for the design stages of new functionality because no real-world data yet exists for that application.
Because threat modeling is a skill intensive exercise, we recommend taking measures to minimize time
investment while maximizing security value:
1. Prioritize by risk - Apply threat modeling first to business-critical applications that would have an
outsize impact on the business if compromised
2. Limit Scope - Perform threat modeling in progressive stages of detail to quickly identify quick wins and
actionable mitigations before spending a lot of manual effort:
a. Star t with simple questions method (See Simple questions method) documented below to
quickly get insight into risks and whether basic protections are in place
b. Progressively evaluate Application Design – as resource and expertise are available, move to
a more advanced analysis using the STRIDE method Advanced threat modeling techniques or
another similar one already used by your team. Start with the architecture level design and
progressively increase detail as time and resources allow:
a. System level design – includes applications and how they interact with each other
b. Application level – includes components of the application and how they interact with
each other
c. Component level – includes how the individual component is composed and how each
element of it interacts with each other
3. Align with Development lifecycle – Optimize your efforts by aligning threat modeling activities with
your application development lifecycles.
a. Waterfall – ensure major projects should include threat modeling during the design process and
during significant updates to the application.
b. DevOps –Trigger threat modeling activities at a frequency that adds security value without over-
burdening the development teams. Good integration points are during the introduction of
significant features or changes to the application and a regular recurring calendar schedule for
example, every quarter for business-critical applications.
c. Legacy applications – These applications typically lack support, source code access, and/or
expertise in the organization, so perform threat modeling on a best effort basis with what
application knowledge/expertise you have available.
Simple questions method
This simple questioning method is designed to get security professionals and developers started on threat
modelling before moving on to a more advanced method like STRIDE or OWASP’s method (see, Top-down
approach through threat modeling).

For each application or component, ask and answer these questions

Are you authenticating connections using Azure AD, TLS (with mutual authentication), or another modern
security protocol approved by your security team? This protects against unauthorized access to the
application and data
 Between users and the application (if applicable)
Between different application components and services (if applicable)
Do you limit which accounts have access to write or modify data in the application to only those required
to do so? This reduces risk of unauthorized data tampering/alteration
Is the application activity logged and fed into a Security Information and Event Management (SIEM) via
Azure Monitor or a similar solution? This helps the security team detect attacks and quickly investigate
them.
Is business-critical data protected with encryption that has been approved by the security team? This
helps protect against unauthorized copying of data while at rest.
Is inbound and outbound network traffic encrypted using TLS? This helps protect against unauthorized
copying of data while in transit.
Is the application protected against Distributed Denial of Service (DDoS) attacks using services like Azure
DDoS protection, Akamai, or similar? This protects against attacks designed to overload the application so
it can’t be used
Does the application store any sign in credentials or keys to access other applications, databases, or
services? This helps identify whether an attack can use your application to attack other systems.
Do the application controls allow you to fulfill security and privacy requirements for the localities you
operate in? (This helps protect user’s private data and avoid compliance fines)

Impor tant: Security is a complex topic and the potential risks are limited only by the imagination of smart
motivated attackers. These questions are designed to help identify readily discoverable gaps that are easily
exploited by attackers. As you develop comfort and competencies with this method, you can look to grow
your ability to threat model by progressing to advanced threat modelling techniques.

Advanced threat modeling techniques

A more comprehensive threat model can identify more potential risks, two popular techniques are STRIDE
and OWASP

Microsoft Security Development Lifecycle has documented a process of threat modeling in and released
a free tool to assist with this process
This method evaluates application components and connections/relationships against potential
risks, which map to the STRIDE mnemonic:
Spoofing
Tampering
Repudiation
Information Disclosure
Denial of Service
Privilege Elevation
This method can be applied to any level of the design from the high level architectural specific
application components.
OWASP – The Open Web Application Security Project (OWASP) has documented a threat modeling
approach for applications, which refers to STRIDE and other methods
 https://www.owasp.org/index.php/Application_Threat_Modeling

Use Web Application Firewalls

Web application firewalls (WAFs) mitigate the risk of an attacker being able to exploit commonly seen security
vulnerabilities for applications. While not perfect, WAFs provide a basic minimum level of security for web
applications.
WAFs are an important mitigation as attackers target web applications for an ingress point into an organization
similar to a client endpoint. WAFs are appropriate for both
Organizations without a strong application security program as it’s a critical safety measure(much like a
parachute in a plane. Note that this shouldn’t be the only planned safety mechanism to reduce the
volume and severity of security bugs in your applications. For details, see Reduce security bug volume
and impact.
Organizations who have invested in application security as WAFs provide a valuable additional defense
in-depth mitigation. WAFs in this case act as a final safety mechanism in case a security bug was missed
by security practices in the development lifecycle.
Microsoft includes WAF capabilities in Azure Application Gateway and many vendors offer these capabilities as
standalone security appliances or as part of next generation firewalls.

Follow best practices for container security

Applications hosted in containers should follow general application best practices as well as some specific
guidelines to manage this new application architecture type
Containerized applications face the same risks as any application and also adds new requirements to securely
the hosting and management of the containerized applications.
Application containers architectures introduced a new layer of abstraction and management tooling (typically
Kubernetes) that have increased developer productivity and adoption of DevOps principles.
While this is an emerging space that is evolving rapidly, several key lessons learned and best practices have
become clear:
Use a Kubernetes managed ser vice instead of installing and managing Kubernetes
Kubernetes is a very complex system and still has a number of default settings that are not secure and
few Kubernetes security experts in the marketplace. While this has been improving in recent years with
each release, there are still a lot of risks that have to be mitigated.
Validate container + container supply chain
Just as you should validate the security of any open-source code added to your applications, you should
also validate containers you add to your applications.
Ensure that the practices applied to building the container are validated against your security
standards like application of security updates, scanning for unwanted code like backdoors and
illicit crypto coin miners, scanning for security vulnerabilities, and application of secure
development practices.
Regularly scan containers for known risks in the container registry, before use, or during use.
Set up registr y of known good containers
This allows developers in your organization to use containers validated by security rapidly with low
friction. Additionally, build a process for developer to request and rapidly get security validation of new
containers to encourage developers to use this process vs. working around it.
 Don’t run containers as root or administrator unless explicitly required
Early versions of containers required root privileges (which makes attacks easier), but this is no longer
required with current versions.
Monitor containers
Ensure you deploy security monitoring tools that are container aware to monitor for anomalous behavior
and enable investigation of incidents.

Next steps
For additional security guidance from Microsoft, see Microsoft security documentation.