Lecture 2

Compute Virtualization:
Bare-Metal (Type 1) Hosted (Type 2)
No Host OS Host OS between VMM and Hardware
Better performance Worse Performance
Hard to develop kernel Easy to implement
Run different Guest OSs Limited in terms of resource access

Characteristics of virtualization:
1. Partitioned
2. Isolated
3. Encapsulated
4. Independent

CPU Virtualization
Full Virtualization Paravirtualization
VM requests are forwarded to VMM Guest OS is modified to be aware of
for binary translation and emulation. virtualization and uses hyper calls to
access hypervisor.
Worse performance due to binary Offers performance similar to non-
translation. virtualized systems.
Supports all types of Guest OSs Only open source OSs that can be
modified.

Memory Virtualization:
VMM stores mapping between guest machine address spaces and physical
machine.
VM OS VMM
Virtual memory Physical memory Machine Memory
Lecture 3
I/O Virtualization
Full Virtualization Paravirtualization I/O through
VMM intercepts the requires each VM to run • Guest can access
request sent by the VM, a frontend driver. When the device directly.
and then sends the real VMs need to access an • Virtually
access request to the I/O device, the VMs send eliminates all
physical device for I/O requests to the emulation and
processing. privileged VM (act as back-end
hypervisor) through the overhead.
Best option for frontend driver, and the • Each device is
correctness and backend driver of the limited to use by
abstraction. privileged VM collects one VM.
the I/O request sent by • Introduces strong
each VM. coupling between
the VM OS needs to be Guest and
modified (usually Linux). hardware.
Specifically, the I/O
request processing
method of the OS needs
to be changed.
Causes severe reduces the performance Best option for
performance loss to the loss of VMM and performance
server. therefore delivers better
I/O performance.
For example, keyboard For example, hard disks
and mouse. and Network Interface
Cards

Mainstream Compute Virtualization Technologies

Open Source Closed Source
KVM Hyper-V
XEN VMware ESXi
KVM vs XEN
Kernel-based VM (KVM) Xen
Type-II Full virtualization. Type-I (Based on Paravirtualization)
Supports both Paravirtualization and
Full virtualization. (Slides 40, 51)

KVM Modes
Guest Mode User Mode Kernel Mode
VMs, including their The quick emulator In this mode, the
CPUs, memory, and (QEMU) typically runs in hardware can be
disks, run in a restricted this mode. QEMU operated. When the
CPU mode. emulates I/O requests. guest OS executes an I/O
operation or privileged
**the simulation instruction, a request
performance delivered needs to be submitted to
by QEMU is low. the user mode, and then
the user mode initiates a
hardware operation
request to the kernel
mode again to operate
the hardware.

The Virtio technology was developed to simply KVM I/O procedure.

Virtio Pros and Cons
Pros Cons
Saves the hardware resources required some old or uncommon devices
for QEMU emulation. cannot use Virtio but can only use
Reduces the number of I/O request QEMU.
paths and improves the performance
of virtualization devices.

Libvirt is an open-source project and is a powerful management tool. It

can manage virtualization platforms such as KVM.
Components of the Xen hypervisor
Backend Frontend
lives in dom0, which provides the within the guest domain, which allows
virtual device. the guest OS to access the virtual
device.
Lecture 4
Types of Network Virtualization
External Virtualization Internal Virtualization
uses switches, adapters, or networks This type of network virtualization
to combine one or more networks into entails creating an emulated network
virtual units. multiple physical within software on a single network
networks are aggregated into one server, enabling VMs to exchange data
software-based administrative entity on a host without using an external
for enhanced efficiency and more network.
practical management. Provides connectivity to hypervisor
kernel.
Resides inside the physical server.
Network adapters, switches, bridges, Consists of virtual switches.
repeaters, routers, and hubs

Functions of Network Virtualization

• It enables the functional grouping of nodes in a virtual network.
• It enables the virtual network to share network resources.
• It allows communication between nodes in a virtual network.
• It enforces routing for communication between virtual networks.

Network Virtualization Pros and Cons

Pros Cons
Improves manageability It needs to manage IT in the abstract.
Lower hardware costs Upfront cost
Improves utilization Lower performance
Rapid scalability

Types of Links in VLAN

Access link Trunk link
Untagged frames. Tagged frames.
Types of Interfaces in VLAN
Access Interface Trunk Interface
Untagged frames. Tagged frames.

Device Virtualization vs Link Virtualization

Device Virtualization Link Virtualization
Divide physical switch into multiple Divide physical link into multiple
logical switches logical links

VLAN Pros and Cons

Pros Cons
Defines broadcast domains Not Scalable
Hardens LAN security Management can become complex.
Improves network robustness
Lecture 6
Layer 2 Switch vs Layer 3 Switch
Layer 2 Switch Layer 3 Switch
Switching operates at Data link layer Switching operates at Network layer
MAC address learning technique used. MAC and IP address learning
technique used.
Single broadcast domain. Multiple broadcast domain.
Devices can only communicate within Devices can communicate within or
the same network. outside the network

Physical Network Devices

Core switches Distribution layer Access switches
switches
very high throughput and are mid-tier speed are the traditional top-
advanced routing switches with an of-rack (TOR) switch that
capabilities. emphasis on uplink regularly consists of 24
speeds. to 48 ports of 1 or
10Gbps speeds

3-Tier Architecture vs 2-Tier Architecture

3-Tier Architecture 2-Tier Architecture
is a good solution but also an is also called Two-layered Hierarchical
expensive solution. Model or Collapsed Core Model. This
architecture is used by small
enterprises that cannot use 3 Tier
architecture.
Small enterprises overcome this by
combining core and distribution layer
into one layer.
High-end layer-3 switch is used as the core of the entire network
because:
• It has a high forwarding performance and can meet the requirements for
forwarding traffic on the entire network.
• The high-end layer-3 switch has a modular structure, with excellent fault
tolerance and high scalability.

Out-of-band management vs in-band management.

Out-of-band management In-band management
If management traffic and service If management traffic and service
traffic are carried on different physical traffic are carried on a same physical
lines and interfaces. channel.

NAT vs Bridged
NAT Bridged
NAT is used for address translation. When the virtual NIC is connected to a
When a VM communicates with an port of the switch, the virtual NIC and
external network the network bridge, with the same IP
through the NAT gateway, the source address configuration, communicate
IP address of the IP packet is with each other in broadcast mode.
translated into the IP address of the
physical network and a record is
produced accordingly.
When the external network accesses
the VM, the NAT gateway forwards
data packets to the
VM based on the record.
The source IP address can be Faster
concealed.

An uplink port is a port on a router or switch designed to connect to

another router or switch or an Internet access device.
Lecture 6
Redundant arrays of independent disks (RAID)
RAID 0 RAID 1 RAID 5 & RAID 6
known as stripping. known as mirror or
mirroring.
When data is stored, writes the same data to similar to RAID 1 but
data is segmented based the mirror disk while they offer less storage
on the number of disks storing the data on the space for data compared
that build the RAID 0 source disk. When the to a RAID 1
volume, and the data is source disk fails, the configuration.
written into the disks in mirror disk takes over
parallel. services from the source
disk.
Fastest, No redundancy best data security among
all RAID levels because
the mirror
disk is used for data
backup.

Storage Architecture for Virtualization

(Storage Area Network (SAN) (Network-Attached Storage
systems) (NAS) systems)
Block-based. File-based.
A SAN is a specialized high‐speed NAS is a single storage device that
network of storage devices and FC serves files over Ethernet and is
switches connected to Host or Server relatively inexpensive and easy to set
via Fiber optic cables. up.
Host uses the storage from storage NAS allows users to directly access file
arrays as if its local to Host. data over the network. In this
configuration,
NAS centrally manages and processes
all shared files on the network.
NAS uses: (CIFS) in Windows server environments, (NFS) protocols for Linux
systems.
From User perspective:

SAN NAS
SAN-connected disks appear to the NAS devices deliver shared storage as
user as local drives. network mounted volumes

Centralized Storage vs. Distributed Storage

Centralized Storage Distributed Storage
all disks are put into disk enclosures It virtualizes all available space
and uniformly managed by the distributed across different hosts into
controller enclosure. The system a single virtual device.
supports dynamic storage capacity The data stored in this virtual storage
expansion and improves fault is also distributed all over the storage
tolerance as well as read and write network.
performance.
Like RAID

Distributed Storage Pros and Cons:

Pros Cons
In terms of capacity expansion, any In addition to the applications running
x86 server with a client installed can on the server, the client software
be a part of the distributed system. installed on the server also consumes
Therefore, this mode delivers great compute resources.
scalability.
Large number of clients adds
complexity to path calculation.
This is why sometimes performance
cannot be linearly improved simply by
adding more clients.
Advantages of Storage Virtualization:
• Easier management.
• Better storage utilization.
• Extend the life of older storage systems.
Lecture 7
Virtual Cluster Features
High Availability Load Balancing Easy Scalability Memory
(HA) Overcommitment
If a node in the It is a cluster all resources are allows VMs to use
cluster becomes technology that pooled to carry more memory
faulty, VMs on distributes loads service VMs. than the
this node is (such as network When the service total physical
automatically services and volume increases, memory of the
migrated to an traffic) to multiple you only need to server by
alternative host network devices. add resources to leveraging specific
that is running Uses the the resource pool, technologies to
properly. intelligent without having to improve VM
scheduling add hardware density.
algorithm to resources. During -Memory
determine the the actual overcommitment
optimal host for implementation, technologies
running VMs. you only need to include:
add servers. • Memory
Ballooning
• Memory
Swapping
• Memory
Sharing.
During High Advantages: Advantages:
Availability (HA), • High • It reduces
the following performance. customers'
issues need to be • Transparency. costs.
addressed: • High reliability. • The feature
• Detecting host helps increase
faults VM density
• Handling VM when the
startup failures memory size of
compute
nodes is fixed.
Memory Sharing Memory Ballooning Memory Swapping
Multiple VMs share the The system automatically External storage is
memory page on which reclaims the unused virtualized into memory
the data content is the memory from a for VMs to use. Data
same. VM and allocates it to that is not used
other VMs to use. temporarily is stored to
external storage.

Virtualization (VMs) Features

Template VM replication (Cloning)
A VM template is a copy of a VM. It In addition to deploying a VM using a
contains VM disks and VM template, you can also quickly deploy
configuration files. a VM using another VM. This function
Creating a VM using a VM template is called VM cloning. Different from
can greatly reduce the time for template-based deployment, VM
configuring a cloning is a process of completely
new VM and installing an OS. copying the source VM at a
specific time point.
IP, Host name are the same, don’t run
them at the same time.

Hot-add VM Console VM Snapshot

is to add compute, are configured with an Snapshot is a fully usable
storage, and network independent copy of a defined
resources to a VM when management interface collection of data that
the VM is started. for users to perform contains an image of the
management operations data as it appeared at
the point in time at
which the copy was
initiated. A snapshot can
be either a copy or a
replication of the
specified data.
The snapshot technology has the following features:
• Snapshots can be quickly generated and used as data sources for
traditional backup and archiving, reducing, or eliminating data backup
windows.
• Snapshots are stored on disks and can be quickly accessed, accelerating
data restoration.
• Disk-based snapshots provide flexible and frequent restoration points for
storage devices.
• Snapshots at different points in time can be used to restore data that is
accidentally erased or damaged online.
Lecture 8

The Internet of Things (IoT) is the concept of connecting everything through the
Internet, including all sorts of communication and interaction: people-people,
people-things, and things-things.
Main IoT Technologies:
• RFID
• Sensors
• Embedded Systems
Big data refers to data sets that are too large or complex to be handled by
traditional data-processing application software.
Characteristics of Big Data:
• Volume: the amount of data
• Variety: the different types of data
• Velocity: the speed at which data is generated
• Value: the value that can be extracted from data.
Artificial intelligence (AI) is an area of computer science that focuses on the
research into and development of theories, methodology, technology, and
application systems for simulating and extending human intelligence with
machines.
Elements of AI:
• Big Data
• Computing Power
• Algorithms
• Scenario
5G refers to the 5th generation of mobile communications technologies and
standards.
5G:
• Enhanced mobile broadband.
• Reliable, low latency.
• Massive number of connections
Containerization is a lightweight virtualization technology that uses OS-level
virtualization. The primary difference between containers and VMs is that each
VM includes the operating system (OS) but a container does not.
Docker is an open-source application container engine. It allows developers to
pack their applications and dependency packages into a portable container, and
later run them on any other Linux machine.
Compared with containers, VMs have the following limitations:
• High resource overheads.
• Redundant operations.
• Slow startup.
Containers benefits are as follows:
• Quick startup.
• Small size.
• Small resource overheads.
OpenStack is a free open-source software platform for cloud computing.
OpenStack requires contributions to be released under the Apache 2.0 license.
OpenStack provides the framework for building a cloud OS. To build a complete
cloud OS, a large number of software components need to be integrated and
work together to provide functionalities and services required by system
administrators and tenants.
OpenStack must integrate with virtualization software to achieve compute
resource pooling of servers.
OpenStack delivers control commands to KVM and performs VM lifecycle
management operations, such as creating, deleting, starting, and stopping VMs.
Cloud Computing Disadvantages:
• Latency.
• Network Congestion.
• Security.
Four steps of data utilization:
• Collect data.
• Cleanse data.
• Associate information.
• Use knowledge.
Fog Computing Edge Computing
Fog computing is an extension of cloud Edge computing creates an open
computing services to the edge of the platform that integrates core
network to decrease latency and capabilities including network,
network congestion. Fog computing compute, storage, and applications at
uses edge devices to carry out a the edge of your
substantial amount of computation, network, to process data where it is
storage, and communication locally, being generated.
instead of storing all data in cloud data
centers.
Fog computing takes place further Edge computing takes place right on
away (in physical distance) from the devices attached to the sensors, or
sensors that generate data. in some cases, on a gateway device
that is physically close to sensors.
• Low latency
• HD video, bandwidth-intensive
• Transmission network efficiency
• High-quality user experience
Microservices Serverless
Microservices are a software Serverless is a cloud-native
development technique that development model that allows
structures an application as a developers to build and run
collection of loosely coupled, fine- applications without having to manage
grained services. servers.
Microservices are an architecture in Serverless is an execution model
which a monolithic application is where the cloud is responsible for
divided into multiple small services executing a piece of code by
such as logging, data search, and dynamically allocating the resources.
more. These services coordinate and And only charging for the amount of
cooperate with each other to provide resources used to run the code.
users with the required functionalities. Serverless computing offerings
Together, these microservices form a typically fall into two groups:
single application. FaaS, BaaS
Pros: To sum up, serverless architecture is
• Independent deployment, designed to allow programmers
flexible scalability focusing on code running without the
• Effective resource isolation complexity of building and maintaining
• Optimized development teams the infrastructure typically associated
with developing and launching an
application
Cons: Serverless computing is a model for
• Complex development and building and managing microservices-
testing process due to project based architectures.
splitting
• Difficulty of ensuring data
consistency between services
Microservices is a way to design an serverless is a way to run an
application. application (or a part of an
application).
Function as a Service (FaaS) Backend as a Service (BaaS)
The service provider provides a The service provider provides a
platform where the developer serverless backend, also known as
develops, runs, and manages their self- backend as a service for the mobile
contained functions without the need application developer. BaaS shifts
to build and maintain an backend functions including data/file
infrastructure. storage, account management, data
collection, operation analysis, message
push, and social media integration to
the cloud, allowing the developer to
focus on their apps’ frontend.