AWS CLOUD COMPUTING UNIT-3

Q) Virtualization of the CPU?

Virtualization of the CPU refers to the process of creating a virtual

version of a physical CPU, allowing multiple operating systems (OS) or
applications to share a single CPU's resources simultaneously. It enables
efficient use of hardware by abstracting the CPU’s capabilities and
distributing them to virtual machines (VMs) or containers. This is
commonly used in cloud computing, data centers, and virtualization
technologies.

How CPU Virtualization Works:

Hardware-Assisted Virtualization:

Modern CPUs from Intel (VT-x) and AMD (AMD-V) include hardware
virtualization support, which allows hypervisors (like VMware, Hyper-V,
or KVM) to create VMs efficiently. This reduces overhead and improves
performance.

Hypervisor Role:

A hypervisor is software that virtualizes the CPU and other hardware

components to manage multiple virtual machines (VMs). It provides VMs
with a virtual CPU (vCPU) that shares the physical CPU.

Time-Slicing:

The hypervisor divides the physical CPU's resources into time slots and
allocates these slots to vCPUs running in virtual machines. This makes it
seem like each VM has its own dedicated CPU.

CPU Emulation:

Some virtualized environments use CPU emulation to simulate CPU

behavior for legacy systems or cross-platform compatibility. This is less
efficient but highly flexible.
Q) Virtualization of Memory?

Virtualization of Memory refers to the abstraction of physical memory

(RAM) by creating a virtualized environment that enables multiple
virtual machines (VMs) or processes to share and manage memory
resources efficiently. This is achieved through a combination of
hardware and software techniques, ensuring that each virtual machine
or application believes it has its own isolated memory space.

How Memory Virtualization Works:

Virtual Memory Abstraction:

The operating system (OS) or hypervisor maps virtual memory addresses

used by applications or VMs to the actual physical memory addresses.
This mapping ensures that each virtualized entity operates in its own
logical memory space, isolated from others.

Memory Management Unit (MMU):

The MMU is a hardware component in the CPU that translates virtual

memory addresses to physical addresses using page tables.

Hypervisor Role:

The hypervisor manages memory resources for all virtual machines. It

allocates portions of the physical memory to each VM and handles
translation between virtual and physical memory.

Paging and Page Tables:

Virtual memory is divided into pages, and physical memory is divided

into frames. The hypervisor or OS maintains page tables to map pages to
frames, enabling efficient memory allocation and access.

Ballooning:

Memory ballooning is a technique used by the hypervisor to reclaim

unused memory from VMs and allocate it to others when needed.
Overcommitment:

The hypervisor can allocate more virtual memory than the physical
memory available by relying on swapping or compression techniques.

Q) Virtualization of I/O Devices?

Virtualization of I/O Devices refers to the process of abstracting and

managing physical I/O devices (like network cards, storage devices, and
USB controllers) in a way that allows multiple virtual machines (VMs) or
containers to share and use these devices as if they were dedicated to
each individual VM.

This enables better resource utilization, flexibility, and isolation in

virtualized environments, such as those found in cloud computing, data
centers, and high-performance computing setups.

How I/O Device Virtualization Works:

Device Abstraction:

The physical I/O devices are abstracted by the hypervisor or OS into

virtual devices. These virtual devices appear to the VMs as if they have
direct access to their own hardware, while in reality, the hypervisor
controls access to the physical devices.

Hypervisor Role:

The hypervisor manages the interaction between VMs and physical I/O
devices. It allocates and virtualizes the hardware resources to ensure
each VM has access to the necessary devices while isolating them from
each other.

I/O Virtualization Techniques:

There are different methods of achieving I/O device virtualization:

Para-virtualization: VMs are aware of the virtualization layer, allowing
the guest OS to optimize I/O operations by working with the hypervisor.

Full virtualization: The guest OS is unaware of the virtualization layer,

with the hypervisor emulating the I/O device.

Direct I/O (Pass-through): A VM can have direct access to a physical I/O

device, bypassing the hypervisor layer (commonly used for devices that
require high performance, like GPUs or high-speed network cards).

I/O Device Emulation:

In full virtualization, the hypervisor emulates the I/O devices to give the
guest OS the illusion that it has direct access to real hardware. For
instance, network or storage controllers are emulated by the hypervisor.

I/O Virtualization Layers:

Virtio (Virtual I/O):

A standardized, high-performance interface for network and block

devices. It is often used in KVM and QEMU environments to virtualize
I/O devices efficiently.

SR-IOV (Single Root I/O Virtualization):

A hardware-based method that enables a single physical I/O device, like

a network interface card (NIC), to present multiple virtual devices to
VMs, improving performance by minimizing the overhead associated
with I/O virtualization.

Q) Virtualization Support and Disaster Recovery

Virtualization plays a significant role in enhancing disaster recovery (DR)

capabilities by providing flexible, scalable, and cost-effective solutions to
recover data and systems in case of unexpected failures.

Role of Virtualization in Disaster Recovery

1. System Abstraction and Portability

Virtualization abstracts hardware dependencies, allowing systems to run

as virtual machines (VMs). VMs are hardware-independent, making it
easy to replicate or migrate them to other physical servers or
environments.

During a disaster, VMs can be quickly restored or migrated to another

server or location without requiring identical hardware configurations.

2. Snapshots and Backup

VM Snapshots: Hypervisors (like VMware, Hyper-V, or KVM) support

snapshots that capture the state of a virtual machine (including memory,
disk, and CPU) at a specific point in time.

Automated Backups: Virtualized environments integrate well with

backup tools, enabling periodic backups of VMs or applications. These
backups can be restored to a different server or data center in case of
failure.

3. Live Migration and Failover

Live Migration: VMs can be moved from one physical host to another
without downtime. This ensures business continuity during planned
maintenance or unexpected hardware issues.

High Availability (HA): Virtualization platforms allow for automatic

failover, where workloads on failed servers are restarted on healthy
servers to minimize downtime.

4. Replication

VM Replication: Virtualized disaster recovery solutions support

asynchronous or synchronous replication of VMs between primary and
secondary sites (data centers).

Example: VMware's vSphere Replication, Microsoft Hyper-V Replica.

In case of a disaster at the primary site, VMs can be quickly activated on
the secondary site.

5. Rapid Recovery and Testing

VMs allow for rapid recovery of systems as entire virtual machines can
be restored much faster compared to reinstalling an operating system,
applications, and configurations on physical hardware.

Virtualization enables non-disruptive testing of disaster recovery plans

using isolated environments. Organizations can simulate disasters and
validate recovery processes without affecting production systems.

Cloud Architecture & Services:

Q) Layered Cloud Architecture: Summary

Layer Role Examples

Hardware infrastructure AWS Data Centers, Azure

Physical Layer
(compute, storage, network) Regions, Google Zones

Virtualization Abstracts hardware into VMware ESXi, Hyper-V, KVM,

Layer virtualized resources Docker, Kubernetes
Resource Orchestrates, provisions, and Terraform, Kubernetes, AWS
Management manages resources CloudFormation
Platform Layer Development and runtime AWS Beanstalk, Azure App
(PaaS) environments Service, Google App Engine
Application Layer Delivers ready-to-use Microsoft 365, Salesforce,
(SaaS) software applications Google Workspace
Web portals, mobile apps, CLI
User Layer User interfaces and APIs
tools
Q) Public, Private and Hybrid Clouds?

Comparison Table: Public, Private, and Hybrid Cloud

Feature Public Cloud Private Cloud Hybrid Cloud
Shared (multi- Dedicated (single- Combination of public
Infrastructure
tenant) tenant) and private
High upfront and Balanced; optimized
Cost Pay-as-you-go
maintenance cost-efficiency
Partial control over
Control Limited Full control
environments
High security for critical
Security Standard security Highly secure
workloads
Limited by
Scalability Highly scalable Scalable via public cloud
infrastructure

Varies (shared Consistent (dedicated Optimized for workload

Performance
resources) resources) type

Managed by Managed by in-house

Maintenance Combination of both
provider IT

Best Use General workloads, Sensitive data, Mixed workloads, DR,

Cases startups compliance flexibility

Definition

A Public Cloud is a cloud environment where computing resources

(servers, storage, networking, etc.) are owned, operated, and managed
by a third-party cloud service provider (CSP).

These resources are shared among multiple customers (multi-tenant

model) and are delivered over the internet.

Public clouds offer cost-effective scalability and are ideal for general
workloads.
Definition

A Private Cloud is a cloud infrastructure dedicated to a single

organization.

It can be hosted on-premises (within an organization's data center) or

managed by a third-party provider but still used exclusively by one
business.

Private clouds provide control, security, and customization for sensitive

data and applications.

Definition

A Hybrid Cloud combines public and private cloud infrastructures,

enabling data and applications to move seamlessly between the two
environments.

This model offers the flexibility to balance cost-efficiency, control, and

security.

Hybrid clouds combine the best of both worlds, ensuring flexibility,

performance, and cost optimization.

Cloud computing models

Q) Comparison of IaaS,Paas and Saas Models

Infrastructure as a Service (IaaS)

Infrastructure as a Service (IaaS) is a cloud computing model that

provides virtualized computing resources—such as servers, storage,
networking, and virtualization—over the internet.

It allows organizations to avoid the expense and complexity of

purchasing, managing, and maintaining physical infrastructure.
Software as a Service (SaaS)

Software as a Service (SaaS) is a cloud computing model where software

applications are delivered over the internet on a subscription basis.

Users can access the software via a web browser without installing or
maintaining it on their local systems.

Platform as a Service (PaaS)

Platform as a Service (PaaS) is a cloud computing model that provides a

development and deployment environment in the cloud.

PaaS delivers hardware, operating systems, middleware, and tools

required for application development, enabling developers to focus on
coding and building applications without managing the underlying
infrastructure.

Comparison of IaaS,Paas and Saas Models

Feature IaaS PaaS SaaS

Application
Infrastructure (VMs, Ready-to-use
Focus development
storage) software
platform

User OS, apps, middleware Application logic and Minimal (just use
Responsibility management code the software)

Scalability High High Moderate

AWS EC2, Azure VMs, AWS Elastic Microsoft 365,

Examples
Google Compute Engine Beanstalk, Heroku Salesforce

Q) Cloud Delivery Models?

Cloud Delivery Models

IaaS: Offers virtualized infrastructure for flexible resource management.

PaaS: Provides a development platform with tools for faster application

development.

SaaS: Delivers ready-to-use software applications with minimal user

involvement.

FaaS: Runs serverless functions, ideal for event-driven workloads.

Q) Cloud Deployment Models?

Deployment Models
Public Cloud: Cost-effective, scalable, and managed by a third-party
provider.
Private Cloud: Highly secure and customizable for single organizations.
Hybrid Cloud: Combines public and private clouds for flexibility and
control.
Community Cloud: Shared infrastructure for organizations with common
interests.

Comparison of Deployment Models

FeaturePubl
ic
CloudPrivat
e
CloudHybri
d
CloudCom
munity
CloudOwne
rshipThird-
party
providerSin
gle
organizatio
nCombinati
on (private
&
public)Shar
ed among
organizatio
nsMulti-
TenancyYe
sNoPartialY
esCostLow
HighModer
ateModerat
e
(shared)Sec
urityModer
ateHighHig
hHighScala
bilityHighL
imitedHigh
ModerateC
ontrolLowH
ighModerat
eModerateB
est
ForGeneral
purposeSec
urity-
sensitive
appsBalanci
ng
flexibilityC
ollaborative
projects
End of Unit-3