UNIT – IV : Networking for Cloud Compu ng

Topic 1: Introduc on

Defini on:

Networking in cloud compu ng refers to the interconnec on of cloud resources, data centers, and
users through the Internet or private networks to deliver cloud services efficiently.

Explana on:

 Cloud compu ng completely depends on network connec vity for data transfer between
users and cloud servers.

 The network acts as the backbone of the cloud, enabling access to services such as storage,
applica ons, and databases from anywhere.

 Without a reliable network, cloud compu ng cannot func on properly because every
opera on — from login to data processing — happens over a network.

Need for Networking in Cloud:

1. Connec vity: Connects cloud servers, storage systems, and users.

2. Resource Sharing: Allows mul ple users to access the same resources at the same me.

3. Scalability: Enables the addi on or removal of resources dynamically.

4. Performance: Ensures high data transfer speed and low latency.

5. Security: Supports firewalls, VPNs, and encryp on for safe communica on.

Key Components of Cloud Networking:

Component Descrip on

Client Device Accesses the cloud services (laptop, mobile, etc.)

Internet / WAN Medium through which data travels

Cloud Data Centers Hosts servers, storage, and networking hardware

Virtual Networks (VPC, VNet) Isolated private networks inside the cloud

Load Balancer Distributes traffic across mul ple servers

Routers & Switches Direct the data packets to the correct des na on

Example:

Drive, your request goes through a network:

Your Device → Internet → Google’s Data Center → Cloud Server → Back to You

This en re process depends on cloud networking.

Advantages of Networking in Cloud:

 High availability and reliability

 Centralized data access

 Cost-effec ve infrastructure

 Fast communica on between virtual machines (VMs)

 Supports distributed applica ons

Conclusion:

Networking is the founda on of cloud compu ng, ensuring smooth communica on between users
and cloud infrastructure.
A well-designed cloud network provides speed, scalability, security, and reliability for cloud services.

Topic 2: Overview of Data Center Environment

Defini on:

A Data Center is a facility that houses a large number of servers, storage systems, networking
equipment, and other compu ng resources used to provide cloud services.

Explana on:

 All cloud services (like Google Drive, AWS EC2, Azure Storage) run inside data centers.

 A data center environment provides power, cooling, network, and security to keep servers
running 24×7.

 It is the physical backbone of the cloud — where user data is processed, stored, and
managed.

Main Components of a Data Center:

Component Descrip on

Servers Handle processing and compu ng tasks.

Storage Systems Store user data and backups.

Networking Devices Routers, switches, and firewalls to connect systems.

Virtualiza on So ware Creates mul ple virtual servers on a single machine.

Cooling Systems Maintain op mal temperature for equipment.

Component Descrip on

Power Supply & UPS Ensure con nuous electricity and backup.

Monitoring & Security Systems Protect against failures and unauthorized access.

Architecture of a Data Center:

A typical data center has three ers (layers):

1. Core Layer: Connects to external networks and handles large data transfers.

2. Aggrega on Layer: Provides rou ng, filtering, and load balancing.

3. Access Layer: Connects individual servers and switches inside the data center.

Characteris cs of Data Centers:

1. High Availability: Redundant systems to prevent down me.

2. Scalability: Can add or remove resources easily.

3. Virtualiza on: Efficient u liza on of hardware resources.

4. Security: Physical and network-level protec on.

5. Automa on: Self-monitoring and self-healing systems.

Types of Data Centers:

1. Enterprise Data Center – Owned by companies for private use.

2. Coloca on Data Center – Rented space in a third-party facility.

3. Cloud Data Center – Owned by cloud providers (AWS, Azure, GCP).

4. Edge Data Center – Located near users for faster access (low latency).

Func ons of Data Centers in Cloud:

 Store and process user data.

 Manage virtual machines and containers.

 Provide backup and disaster recovery.

 Support distributed compu ng and large-scale storage.

 Deliver cloud applica ons to end users.

Conclusion:

A Data Center Environment is the founda on of cloud infrastructure.

It ensures con nuous availability, scalability, and performance of cloud services by integra ng
servers, networks, storage, and virtualiza on efficiently.

Topic 3: Networking Issues in Data Centers

Introduc on:

A data center has thousands of servers all connected through high-speed networks.
When these servers communicate or share data, several networking issues can occur — affec ng
speed, reliability, and performance.

Main Networking Issues:

Latency

 Meaning: Time delay between sending and receiving data.

 Example: When you click “upload” and it takes a few seconds to start, that delay = latency.

 Cause: Long distances, slow rou ng, or overloaded switches.

 Solu on: Use faster connec ons (fiber), and place servers closer to users (edge data
centers).

Remember: Low latency = faster cloud performance.

Bandwidth

 Meaning: The amount of data that can be transmi ed per second.

 Example: A 100 Mbps link can send 100 megabits every second.

 Issue: If bandwidth is low, data transfer becomes slow — especially when many users are
ac ve.

 Solu on: Use high-capacity links and balance traffic.

Conges on

 Meaning: When too much data travels through the network at once, like traffic jam on a
road.

 Effect: Packet loss and delays.

 Cause: Many users or processes sending large data simultaneously.

 Solu on: Use smart rou ng and conges on control protocols (like DCTCP).
 Scalability

 Meaning: The ability to expand the data center network easily as demand grows.

 Issue: Adding more servers increases network complexity.

 Effect: Slower performance if not managed properly.

 Solu on: Use scalable architectures (like spine-leaf topology) that handle growth easily.

Fault Tolerance

 Meaning: The ability of the network to con nue working even if some components fail.

 Example: If one cable or switch fails, another path should take over automa cally.

 Solu on: Use redundant (backup) connec ons and automa c failover systems.

Security (some mes added as extra point)

 Issue: Unauthorized access or data the while transferring over the network.

 Solu on: Use firewalls, encryp on, and network monitoring.

Summary Table:

Issue Meaning Example / Solu on

Latency Delay in data transfer Place servers closer to users

Bandwidth Data capacity Increase link speed

Conges on Too much traffic Use DCTCP protocol

Scalability Growth handling Use scalable topology

Fault Tolerance Backup on failure Redundant paths

Security Protect data Encryp on, firewalls

Topic 4: Transport Layer Issues in DCNs (Data Center Networks)

What Is the Transport Layer?

In networking, data travels in layers (called the OSI Model).

The Transport Layer is the 4th layer, and it controls:

 How data is broken into packets

 How fast it is sent

  Whether it reaches safely or gets lost

The most common transport layer protocol is TCP (Transmission Control Protocol).

How It Works in Cloud Data Centers

Inside a data center, thousands of servers keep sending and receiving small packets of data to each
other.
Example: One server stores files, another processes them — they constantly exchange info.
TCP handles all this communica on.

But here’s the problem

Problems with TCP in Data Centers

TCP was originally designed for the internet, not for high-speed, short-distance data center
networks.
So when it’s used in data centers, several issues occur:

Problem Explana on

Conges on Too many packets at once cause delays.

Packet Loss TCP sends too much data before checking if the path is free.

Latency TCP reacts slowly to conges on; response me increases.

Unfair Bandwidth Usage Some servers get more bandwidth than others.

CPU Overhead TCP consumes more CPU resources for managing data.

In short: TCP = reliable but too slow for massive, fast-moving data centers.

Solu ons: Improved Protocols

A. DCTCP (Data Center TCP)

 A modified version of TCP specially made for data centers.

 It detects conges on early using a feature called Explicit Conges on No fica on (ECN).

 Reduces data packet delay and improves performance.

 Provides fair bandwidth sharing among servers.

Think of it like a smart traffic system that slows cars before a jam happens.

B. RDMA (Remote Direct Memory Access)

 Allows one server to directly access another server’s memory without involving the CPU.

 This reduces processing me and speeds up data transfer.

 Used in high-performance compu ng and modern cloud systems.

Example: Instead of sending a message to ask “give me this data,” it just picks it directly from
memory — super fast.
 Summary Table

Protocol / Issue Descrip on Result

TCP Default protocol, reliable but slow Causes latency, conges on

DCTCP Modified TCP for DCNs Handles conges on efficiently

RDMA Direct memory access Ultra-fast data transfer

Topic 5: Cloud Service Providers

Who Are Cloud Service Providers (CSPs)?

Cloud Service Providers are companies that offer cloud compu ng services — like storage,
networking, databases, and compu ng power — through the internet instead of you needing
physical servers.

Examples of Major CSPs

Provider Full Form Key Product/Service Headquarters

Amazon Web
AWS EC2, S3, Lambda, RDS USA
Services

Virtual Machines, Blob Storage,

Microso Azure — USA
Azure SQL

Google Cloud Pla orm Compute Engine, BigQuery, Cloud

— USA
(GCP) Func ons

IBM Cloud — Cloud Foundry, Watson AI USA

Oracle Cloud Infrastructure

— Autonomous DB, Compute, Storage USA
(OCI)

Types of Services Offered (3 Main Categories)

Service Type Meaning Example

IaaS (Infrastructure as a
Rent servers, storage, and networking AWS EC2, Azure VM
Service)

Provides a full pla orm to build and Google App Engine, Azure App
PaaS (Pla orm as a Service)
deploy apps Service

SaaS (So ware as a Service) Ready-to-use applica ons over internet Gmail, Zoom, Salesforce
 Comparison of Major Providers

Feature AWS Azure GCP

Launch Year 2006 2010 2008

Market Share ~32% ~23% ~11%

Strength Largest variety of services Strong in enterprise integra on Best in data analy cs & AI

Pricing Pay-as-you-go Hybrid op ons Flexible billing

Popular Users Ne lix, Airbnb LinkedIn, Adobe Spo fy, Snapchat

Why Companies Choose CSPs

1. Scalability – Increase or decrease resources easily

2. Cost Efficiency – Pay only for what you use

3. Global Reach – Data centers worldwide

4. Security – Advanced encryp on and access control

5. High Availability – Backup, load balancing, and redundancy

Exam Point / Short Notes:

Cloud Service Providers (CSPs) offer IaaS, PaaS, and SaaS through data centers worldwide. Major
CSPs include AWS, Azure, GCP, IBM Cloud, and Oracle Cloud. They provide scalable, secure, and cost-
effec ve solu ons for businesses.

Quick Summary Table

Term Meaning Example

AWS,
CSP Cloud Service Provider
Azure, GCP

EC2, Azure
IaaS Infrastructure as a Service
VM

App
PaaS Pla orm as a Service
Engine

Gmail,
SaaS So ware as a Service
Zoom