100% found this document useful (5 votes)
1K views3 pages

Architecting Microsoft Azure Solutions

Azure Container Instances (ACI) provides a serverless compute service for short-running container workloads using Docker containers and an HTTP-based API. Azure Batch is suited for scientific calculations run as batch jobs using virtual machine scale sets (VMSS) and HPC clusters. Azure Kubernetes Service (AKS) provides orchestration and load balancing for long-running container workloads.

Uploaded by

mmihm
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
100% found this document useful (5 votes)
1K views3 pages

Architecting Microsoft Azure Solutions

Azure Container Instances (ACI) provides a serverless compute service for short-running container workloads using Docker containers and an HTTP-based API. Azure Batch is suited for scientific calculations run as batch jobs using virtual machine scale sets (VMSS) and HPC clusters. Azure Kubernetes Service (AKS) provides orchestration and load balancing for long-running container workloads.

Uploaded by

mmihm
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

HPC Workload Series Free and Shared Basic Serverless compute service Azure Container Instances (ACI)

Availability Set Basic • One ACI = one Docker container


• 2 fault domains for classic
• A8-11 General purpose • Ideal for development, testing, and non-critical work Event-driven actions and triggers
• N Graphic GPU optimised • Up to 3 instances (manual) • No SLA • HTTP-based API endpoints (HTTP triggers) • Role Based Access Control (RBAC)
• 3 fault domains for Resource Manager deployments Standard • Short-running workloads
• H High performance computing Standard • Timer triggers
• 5 update domains Up to 10 instances (auto scale) Azure Container Services (AKS)
Scale Set HPC Pack • • Ideal for production and cost effective Programming Languages
• Windows Server 2012, 2016, and Linux • 5 Slots • Data replication between two nodes • C#, F#, Node.js, Java, PHP, PowerShell, Batch, JavaScript, • Load balancing
• Max 100 VMs • Daily backups • Orchestration
• Create HPC clusters on-prem • High availability SLA Python, Typescript
• Max 1000 VMs with placement groups (auto scale) • Azure Traffic Manager • Long running workloads
Azure Batch Premium Plans
• Managed disks needed for large scale sets Premium
• Most cost-effective option for scientific calculations • Redis persistence • Consumption App Service Plan (cost effective)
VM Series Cloud-native HPC solution • Up to 20 instances (auto scale)
• A0-7, Av2, B General purpose • Create workloads > 53GB • Other App Service Plans
• HPC head node and compute nodes • 20 Slots • Ability to isolate
• F Compute optimised • Daily backups
• Virtual Machine Scale Sets (VMSS)
• D,E,G Memory optimised • Azure Traffic Manager
• VMs using RDMA are placed in same VMSS
• L Storage optimised Isolated
• Virtual Network
• N Graphic GPU optimised
• Azure Blob Storage for node disks • App Service Environment (ASE) – scalable, secure Cloud Infrastructure Ready
• H High performance computing • Up to 100 instances/plan or 100 plans with one instance • Host on VMs as-is
Join VMs to domain Hybrid HPC solution
+ ExpressRoute to connect cloud with on-prem Service that exposes different apps as APIs Cloud DevOps Ready
• Enable Azure AD Domain Services • Use containers to develop and deploy
+ VPN Gateway endpoint between cloud and on-prem API Gateway • Decouple application from infrastructure
• Bridge between app and outside world Cloud Optimised
• Enhanced security, policies, authentication
• Modernise mission critical application
• Caching, throttling
Routing methods API Management Portal
• Performance, Weighted, Priority, Geographic • Define custom APIs
Handle load & locate closest geo region at DNS level • Package APIs into open or protected products
Developer Portal
• Developers can access APIs and documentation
Up Orchestration Platform
• Select different (better) Service Plan • Cloud and on-prem
Out • Container orchestration
Hybrid Connections Data-integration service • Scale out Web App manually or automatically Lifecycle Management
• Establish a rendezvous point in the cloud • Create workflows to automate data move + transform • Service developer (creates microservices)
• On-prem app connects using HTTP/ Sockets to cloud • Connect to ML, HDInsight, Data Lake Analytics Workflow Driven • Application developer (creates applications)
WCF Relays (Service Bus Relays) • Data sent over HTTP using certificates Integration with cloud and on-prem services • Application administrator (creates config & packages)
• On-prem app uses WCG bindings to connect to Srv Bus • No firewall ports need to be opened • BizTalk, … • Operator (deploys, monitors, maintains)
Multiple programming languages
• ASP.NET, Core, Angular, React.js, Java, Python
Securing Web API
• Azure AD
• Azure AD B2C – with Facebook and Google providers
Enables access from app to other services • Active Directory Federated Services (ADFS)
Connects Azure and on-prem applications using TCP Cache static content to multiple regions • API Management – policies, API keys, throttling, …
• Deploy app inside a VNet
• Uses Azure Relay Service
• Access services within same VNet (VMs, DBs, …)
• Part of App Service and is a separate Azure feature
• TCP or UDP

Access on-prem web apps from the cloud





Provides single sign on (SSO) + secure remote access
Connector – lightweight agent on on-prem server
External endpoint – direct URL or access via MyApps
Bridge between on-prem data sources and Azure
• Uses Service Bus
• Azure -> Analytics, Logic Apps, Flow, Power Apps, …
• On-Prem -> SQL Server, SQL Analytics, SharePoint, …
Architecting Microsoft Azure Solutions 1

Provides central repository Cloud Service for big data processing and analytics General-purpose v1 Unstructured data – VHDs, images, audio, etc. VNets Load Balancing
• One catalog per tenant • Data pipelines, activities, datasets, linked services, triggers, • Classic, does not support latest features. • Max 1TB page blob, 200GB block blob • Max 50 VNets per subscription • Transport Layer 4
• Sources – Blob Storage, Data Lake, QL Server, Oracle, … pipeline ru, parameters, control flow General-purpose v2 Access tiers Subnets • Any protocol
• Available in - East US, East US2, West Europe • Newest, that combines v1 and blob storage • Hot – optimised for frequently accessed data • Max 1000 subnets per VNet • Azure VMs and Cloud service endpoints
• Latest features at a reduction in costs • Cool – Suitable for backups and not often viewed data • Max 10 VNet connections (peering) per subscription • VNet: Internet and internal facing
Blob storage • Archive – set at blob level, cannot be read or modified Pubic Address • Endpoint monitoring: Supported via probes
• Same features as storage v2 acc, but only block blobs. • Max 60 public dynamic addresses per subscription Types
Replication (X redundant storage) • Max 20 public static addresses per subscription • Basic
• Locally – 3 copies within data center Private Address • Standard … up to 1000 VMs, HA ports, and NSG.
• Zone – US East 2 and US Central, 3 datacenter copies • Max 4096 private addresses per VNet
• Geo – three regional copies
DNS
Big data storage and analytics service • DNS for multiple VNets requires own DNS server
• Based on Hadoop Yes Another Resource Negotiator (YARN)
• Solutions - Store, Analytics, and HDInsights Semi-structured, non-relational data
Massive Parallel Processing (MPP) Data Lake Store • Suitable for datasets without complex joins
• Uses Hadoop/Spark and Machine Learning for insights • Access via OData and LINA queries
• Storage repository for big data workloads
• Uses Data Movement Service (DMS) between nodes • Max 500TB data Gateway
• Unlimited structured, semi-, and unstructured data
Data Lake Analytics • DNS level
• Uses serverless approach • Application level 7
• Pas-as-you-go, monthly commitment Create file shares in the cloud • HTTP and HTTPS
• Uses U-SQL to analyse the data • Access with Server Massage Block (SMB) protocol Traffic management • VNet: Any public or internal IP address
• Cached fast access on Win Server using Azure File Sync • DNS level • Endpoint monitoring: Supported via probes
HDInsights • Any protocol
• Deploys Hadoop components in form of clusters in cloud SSL
• VMs, Cloud Service, Web Apps, and external endpoints • SSL off loading to avoid costly decryption
• Opensource service for analysing and processing data • VNet: Internet facing
• Apache Hadoop, Spark, HBase, Storm, Kafka, Interactive Q Asynchronous processing of messages Firewall
• Endpoint monitoring: HTTP/HTTPS GET
• Microsoft R Server • REST.API supports GET, PUT, and PEEK • Web Application Firewall (WAF)
• Messages max 64KB and max 7days lifetime
Load balancing
• Use with load balancer for high-avail and high-per
Same architecture as SQL Server Analysis
• Enterprise grade data modelling in the cloud
Integrated storage spanning on-rem an cloud
• iSCSI and SMB support
StorSimple Virtual Array
Open source relational database • Hyper-V 2000 R2 and VMWare 5.5
Azure VPN
• Used by PHP developers, CMS WordPress • iSCSI server (AN) or File Server (NAS).
• Basic – max 10 site-site, 128 point-site, avg 100Mbps
• ACID, replication, Performance, security, extensibility, StorSimple 8000 Series Used for VMs stored in Az Blob storage as page blobs. DMZ • VpnGw1 – max 30 site-site, 128 point-site, avg 650Mbps
concurrency, JSON support • Leased physical device • Standard – unmanaged HDD disk drives. LRS and GRS • Network Security Groups (NSG) • VpnGw2 – max 30 site-site, 128 point-site, avg 1Gbps
• Pricing • Virtual Appliance Manager replicates data to cloud redundancy only. • User Defined Routes (UDR) • VpnGw3 – max 30 site-site, 128 point-site, avg 1.25Gbps
• Basic – 1TB, 4 CPUs, locally redundancy • Premium – SDD, high-performance disk support • Firewalls
• General Purpose – 1TB, 4 CPUs, local+geo redundancy Site-to-site
Network Security Groups • Requires Routing and Remote Access Service (RRAS)
Relational database • Memory Optimised – 1TB, 5 CPUs, local+geo red. • Inbound and outbound rules
• Elastic Database Pools (eDTUs) • Internet Protocol Security (IPSec) connection
• Checked between VMs, VNets, and other services • Internet Key Exchange (IKE) management protocol
• Individual databases (DTUs) Applied to one or more subnets or network interfaces
• High availability, geo-replication, failover groups
• Point-to-site
• Low order numbers are higher priority • Connect IKE2 or Secure Socket Tunneling Protocol (SSTP)
• Backup and Recovery
• Basic – 7 days retention
User Defined Rules • No RRAS device required
• Create UDRs & IP forwarding by creating a routing table VNet-to-Vnet
• Standard and Premium – 35 days
• Restore - Point-in-time, deleted DB, Geo, and Az Rich search experience over Azure storage Virtual Network Service Tunneling • Max 10 VNet connections (peering) per subscription
Premium Azure Table Storage • Force external traffic through a site-to-site VPN tunnel
Recovery Vault • Multi-model and globally distributed database • SQL Database, CosmosDB, Blob Storage ExpressRoute
SQL Server Stretch Database Open source relational database • Low latency, high availability, high performance • Text search, analysis, and linguistic analysis Web Application Firewall • Any-to-Ant (IPVPN) – provider sets up secure connection
• Move or archive cold data from on-premises SQL Server to • Open Source, ACID, Replication, Performance, Security, Tiers • Part of Application Gateway and based on OWASP 3.0 • Point-to-Point Ethernet –two provider connections
Concurrency, JSON, JSON Indexing, Extensibility APIs • Can protect max 20 applications behind an App G/W
Azure SQL • SQL, MongoDB, Gremlin (Graph), Table, Cassandra • Free, Basic • Co-Located at Cloud Exchange – two cross connections
• Standard S125GB,50 indexes, S2100,200/S3200GB/HD1000 indexes • Examples: SQL Injection, Cross-Site Scripting, Bots, … • Maximum 10GB

AJATO Transformations Limited | 2018.09 | github.com/wpschaub/Quick-Reference-Posters


Directory and identity management Synchronise on-prem AD identities with Azure Store cryptographic keys and secrets Security for data in transit, rest, and in use Define and enforce rules and actions for resources
• Plans – Free (no SLA, 500k objects), Basic, Premium P1/P2 • AAD password hash synchronisation
Implement the principle of least permissions
• Service Tiers: Standard and Premium • HTTPS – security in transit • Roles in Azure can be added to a scope • NOT about users, groups, or application access
• Protocols – OAuth 2.0, OpenID Connect • User passwords hashes synched between AD and AAD • Hardware Security Modules (HSM) with Premium • Transparent Data Encryption – security at rest • Apply governance strategy
• Scope cam be subscription, Resource Group, or Web App
• Endpoint V1 • Hash synched with any change • Always Encrypted – data in use, AlwaysEncrypted columns • Example: All VMs use managed disks
• Set 2000 role assignments from Portal, PS, CLI, Rest API
• Work and school accounts • Provides single sign-on (SSO)
• Built-in Roles: Owner, Reader, Contributor
• Azure Active Directory Library (ADAL) • AAD pass-through authentication
• Endpoint V2 • Passwords are not synchronised, but validated on-prem
• Work, school, and personal accounts • Provides single sign-on (SSO)
• Microsoft Authentication Library (MSAL)
Microsoft Graph
• Connects multi services and provides single endpoint
• AAD is integrated in Microsoft Graph Premium protection for Az identities
• Detect identity based issues
Manage and control access inside an Az AD tenant • Detect compromised identities
• Az AD Prem P2 or Enterprise Mobility + Security E5 feature • Policies: MFA registration, user risk, sign-in risk
Cloud identity management for mobile and web apps • Grant permanent or temporary role access
• Leveraged using MSAL Encrypt Windows and Linux VMs • Flow: User request, review, approval, notification, action,
• Social Accounts – Facebook, Google, LinkedInn • Windows – Bitlocker monitor
Authentication provider for external users to on-prem • Enterprise Accounts – OpenID Connect, SAM • Linux – dm-crypt
• WEB SSO for federated users accessing on-prem apps, • Local accounts – email/user and password
using Azure AD Connect • App must be registered inside Azure B2C tenant
• Web Services (WS) – WS-Federation compatible
• No external user account management – own credentials Advanced Thread Protection and Security Mngt.
using Security Assertion Markup Language (SAML) • Features:
• Install on-prem of Azure VM and use MS Graph. • Centralised policy management
Hybrid cloud and data management tool • Continuous security assessment
• Manage on-prem and Az infrastructure • Actionable recommendations
• Azure, AWS, Win Server, Linux, VMWare and OpenStack • Advanced Cloud protection
Enables organizations to work safely with others • Services: • Prioritised alerts and incidents
• Enabled by default for all AAD tenants • Security and Compliance Solution • Integrated security solutions
• Integrated with Office 365 • Security and Audit • Tiers: Free and Standard (hybrid environments)
Two step verification (MFA) • AD Premium Features requires license ration of 5:1 Encryption for data at rest • Security Domains • Advanced Threat Detection
Every AS Premium licence = five external users Managed identity for resources in Azure • Storage Service Encryption (SSE) • Notable Issues • Activity group, campaign, and threat summary report
• Know – password
• Set conditions for users, for example, enforce MFA • Service Principal only known within bounds of Az resources • Written to storage account using 256-bit AES encryption • Detection
• Have – phone, verification app, 3rd party OAuth tokens • Az Endpoint Protection
• Use policies to delegate permissions • Assign appropriate Role-based Access Control (RBAC) • Set with Portal, PowerShell. CLI, and REST API • Thread Intelligence
• Are - biometrics • Anti malware protection for Az and on-prem VMw

Automation ensures consistency and saves time


• Development, testing, acceptance, and production
• PowerShell – create resources and configure
• Desired State Configuration (DSC) – enforce config
• Features: Configurations, Resources, Local Config Mgr.
Meet performance and SLA requirements
• Vertical scaling – change VM sizes
• Horizontal Scaling – add / resource resources
• Strategies
• Monitoring and alerting
Architecting Microsoft Azure Solutions 2
• Azure Automation
• Process Automation – automate management • Decision Making Logic – automation runbooks
• Configuration Management – DSC, PowerShell • Az Monitoring Scale – integrated in Az Monitor
• Update Management – Cloud + on-prem environments • App Architectures – Service Fabric scales
• Shared capabilities horizontally
• 3rd Party
• Chef – virtual and physical config management, Windows
+ Linux + Mac
• Puppet
• Azure Event Grid – supports automation tasks
• Azure Logic Apps – supports call to automation runbooks
• Azure DevOps – CI/CD

Create modern, intelligent applications, with AI/ML Environment to build and deploy bots
• Artificial Intelligence (AI) & Machine Learning (ML) • Freeform communication
• Services: Vision, Speech, Language, Knowledge, Search • Tiers
• Vision – Categorise, moderate, classify, index, … images • Free – up to 10,000 messages
• Speech – Speech enabled, recognition, translate • Standard S1 – pay for 1,000 messages at a time, SLA
• Language – LUIS, spelling, linguistic, text analysis, web, …
• KB – Personal experience, train AI to converse naturally
• Search – Bing, autosuggest, entity and custom search

Internet of Things (IoT) Hub


• Send massive amounts of data to Az for processing
Monitoring solution in Az Portal Algorithms to apply complex math calc to big data • Bi-directional, secure and routable communication
Collects and analyzes log files from resources • Tools • Scale up to millions of connected devices
• Azure and on-prem resources • Infrastructure metrics and logs for Az services
• Capabilities: • Machine Learning Studio – drag/drop predictive models • Integrated with Azure Monitor
• Analysis tools – OMS, Security Center, AI, PowerBI • Leaning Workbench – end-end science solution • Tiers: Basic (8k msg/day), S1 (400k), S2 (6M), S3 (300M)
Asynchronous processing of messages • Activity Log – info on all types of events
Reliable, brokered messaging system • Diagnostics Settings – info on events within specific srv • AI Gallery – community-driven solutions Event Hub
• Messages up to 64KB in size • Ideal for Integration and IoT scenarios • ML Modules – out-box models for analyzing data
• Metrics – time-based metric points for resources • Ingress of device data streams
• 7 days retention maximum • Messages up to 256KB (basic) and 1MB (premium) • Data Science VMs – preconfigured workloads
• Alerts – View and manage Az alerts • One-way communication
• Messages become visible after 30sec if not deleted • Queues – first in first out (FIFO), one consumer • Aggregated metrics monitoring
• Multiple receivers • Sessions – grouping of messages by session ID • Tiers: Basic (100 connect), Standard (1K), Dedicated (25K)
• Topics – Publish/subscribe by multiple consumers IoT Edge
• Subscriptions – Apps connect to sub to get to topics • Installed at the edge of on-prem network, DMZ
• WCF Relays – gateway for on-prem WCF services to Azure • Collect device data and send to IoT Hub
• Tiers – Basic, Standard (topics, tx, sessions), Premium Helps you follow best practices for Az deployments
• High Availability Pipeline for event processing and real-time analysis
• Security
Az Portal Dashboard showing resource issues • Sources – Apps, sensors, IoT Hub, Event Hub, Blog storage
• Performance • Views • Targets – Data Lake, PowerBI, SQL data Warehouse
• Costs • Service issues
• Planned Maintenance
• Resource Health
• Health Alerts
Queues - Standard queuing with messages up to 64KB
Brokering at enterprise scale with messages up to 1MB, transactions, and sessions - Service Bus Provide valuable insights into IoT data
• DB storage for massive amounts of data
Secure and high-quality streaming and storage • Sources – IoT Hub, Event Hub
• Flow – Upload → Encode → Secure → Analyse
• Join data – metadata, telemetry, and visualise
• Cognitive Azure Media Analytics
• Features – Integration, storage, visualization, query
• Indexer, Hyper lapse, Motion detect, summarize,
Az resource network monitoring for network comms Monitoring solution for cross-platform apps character recognition, face recognition, and moderation
• Capabilities • Az and on-prem apps
• Topology • Events
Push notifications from backends to mobile • IP flow velocity • Rate data
Event management across Azure resources • Scenarios – Send codes, notifications, news • Next Hop • Exceptions
• Apps are notified when an event happens • Tiers • Security Group View • Page views and performance
• Throughput of millions of events and 24h retry • Free – 1 million messages / month • VPN diagnostics • Diagnostic logs
• Publishers – Az subscriptions, Event Hubs, Topics, IoT Hub, • Basic – 10 million messages / month • Packet Capture • Custom Events IoT Hub – Two-way communication
Resource Groups, Blob storage, Service Bus, V2 storage, … • Standards – 10 million messages / month • Connection Troubleshooting • Integration One-way communication for cost effective data ingest – Event Hub

AJATO Transformations Limited | 2018.09 | github.com/wpschaub/Quick-Reference-Posters


Architecting Microsoft Azure Solutions 3

DC/OS, Docker Swarm,


Kubernetes )

AJATO Transformations Limited | 2018.09 | github.com/wpschaub/Quick-Reference-Posters


AJATO Transformations Limited | 2018.09 | github.com/wpschaub/Quick-Reference-Posters

You might also like