Understanding Cloud Infrastructure - I

Course: Cloud Infrastructure

Lecture Topic: Understanding Cloud Infrastructure - I
Instructor: Ab Rouf Khan
(https://www.linkedin.com/in/rouf-khan/)
 TODAY’S AGENDA

What is cloud computing?

Advantages of integrating cloud services with
traditional IT solutions
Architecture of an application deployed on cloud
Cloud service and deployment models
Fog computing
Cloud support for blockchain, artificial intelligence,
analytics, software development
Introduction to Cloud Computing
 INTRODUCTION TO CLOUD COMPUTING
Traditional Data Centre - Email Services

Requirements for setting up an email service Application-specific installation

for a company:
Server hardware
Server operating system (OS)
Installation of device-specific applications
Operating system
Example: Microsoft Exchange

Challenges:
Server hardware failure
Operating system failure Hardware
 INTRODUCTION TO CLOUD COMPUTING
Traditional Data Centre - Schools9

Schools9 is an education portal

designed to display exam results.

Challenges:
Increase in the web traffic on exam
result days

Possible Solution:
Purchase of large servers to handle
the web traffic on result days.
But, is it an efficient solution?
 INTRODUCTION TO CLOUD COMPUTING
Data Centre

Data Centre
This is a physical facility that companies use to house hardware components that store data and applications.

Key Components:
Switches
Routers
Firewalls
Storage systems
Servers
Application-delivery controllers

The Key Components Provide:

Network Infrastructure - Switches, routers, firewalls
Storage Infrastructure - Storage systems
Computing Resources - Servers, application-delivery controllers
 INTRODUCTION TO CLOUD COMPUTING
Traditional Data Centre

Traditional Data
Centres

Colocation On-Premises

Third-Part
Organisation Organisation
y

Servers, Set up, Servers

Maintenance Set up Maintenance
storage , storage, etc.
 INTRODUCTION TO CLOUD COMPUTING
Traditional Data Centre

Types of Data Centres:

On-premise:
● The organisation sets up and maintains the data centre.
● The organisation owns all the equipment.
Colocation:
● The organisation ties up with a third-party company to set up and maintain its data centres.
● The organisation provides the required computer servers, storage and networking.
● A third-party company provides power, cooling and physical security.

In some cases, the network service provider manages the networking between the data centre and
the organisation.
● Example: Email server (on-premise data centre or a colocation data centre)
 INTRODUCTION TO CLOUD COMPUTING
Problems With Traditional Data Centres

High Set-Up Cost

● It requires buying all the software and hardware to set up the system.
High Maintenance Cost
● Dedicated resources are required for monitoring and maintaining of
the data.
Low Scalability
● The changing demands of the business require more computing and
storage resources.
The traditional data centre will not be scalable in such scenarios.
INTRODUCTION TO CLOUD COMPUTING
What is Cloud?

What do you think a Cloud is?

 INTRODUCTION TO CLOUD COMPUTING
What is Cloud Computing?

Cloud computing is the outsourcing of data storage, management

and processing tasks to a third-party service provider.
A third-party service provider (or cloud service provider):
● Stores the data
● Computes on behalf of the user
● Manages the distributed system of hardware and software
Examples:
● Data storage in Google Drive
● Video streaming on YouTube
● File hosting services on Dropbox
● Web hosting for WordPress
 INTRODUCTION TO CLOUD COMPUTING
Features of Cloud Computing

01 Pay-as-you-go / economical

02 On-demand self-service

03 Access control

04 Measured service

05 Resource pooling
 INTRODUCTION TO CLOUD COMPUTING
Features of Cloud Computing

06 Easy maintenance

07 Automated service

08 High availability

09 Resilience

10 Fault tolerance
 INTRODUCTION TO CLOUD COMPUTING
Features of Cloud Computing

11 Scalability

12 Rapid elasticity

13 Security

14 Broad network access

15 Easy migration
 INTRODUCTION TO CLOUD COMPUTING
Traditional Data Centre (On-Premise Systems) vs Cloud-Based Data Centre

PROBLEMS - Traditional IT Data Centre SOLUTIONS - Cloud-Based Data Centre

Non-availability of hardware when required On-demand self-service

Resource pooling

Firewall failure at points of high traffic Easy maintenance

Cannot replace or return the unused component Economical

On-demand service
Pay-as-you-go

Slow speed for low server bandwidth High availability

Unavailable service / network issues Fault tolerance
Resilience - provides backup and recovery
 INTRODUCTION TO CLOUD COMPUTING
Traditional Data Centre vs Cloud-Based Data Centre

Traditional IT Data Centre Cloud-Based Data Centre

Difficult to migrate the small changes in a short Easy migration

time

User access not granted without permission Access control

Not accessible Broad network access
Automated service
Security

Services are not incrementable Scalability

Cannot be decremented based on the Rapid elasticity
requirements on runtime
 INTRODUCTION TO CLOUD COMPUTING
Traditional Data Centre vs Cloud-Based Data Centre

Traditional IT Data Centre Cloud-Based Data Centre

High infrastructure cost Easy migration

High infrastructure cost despite less usage Measured service

Elasticity - able to adapt to workload changes
Pay-as-you-go

Centralised infrastructure More secure

Any lapses can compromise the whole system
security
 INTRODUCTION TO CLOUD COMPUTING
Enabling Technologies

Grid Resilience
Computing Computing

Virtual Utility
Cloud Computing
Computing Computing

Multitenancy Data Centre

Technology Technology
 INTRODUCTION TO CLOUD COMPUTING
Enabling Technologies - Feature Mapping

Grid Computing:
● A group of computers work together as a single system, enabling increased capability
● Higher scalability
● Rapid elasticity
Virtual Computing:
● Allows users to ‘virtually’ use systems that are geographically located at another place
● Resource pooling
Resilience Computing:
● Ensures other systems remain serviceable at all times
● Ensures availability despite failure of a part
● Improves resilience
INTRODUCTION TO CLOUD COMPUTING
Enabling Technologies - Feature Mapping

Utility Computing:
● Computing and storage resources provisioned as ‘utility’
● Specific usage charges
● Pay-as-you-go
● Measured service model
● Cost effective
Data Centre Technology:
● Setting up, configuring and maintaining data centres
● Automated service
● Easy maintenance
● Easy migration
Multitenancy Technology:
● Multiple users utilise the same resources
● Ensure on-demand self-service
● Broad network access
 INTRODUCTION TO CLOUD COMPUTING
Evolution of Cloud Computing

1960s 1990s 1994 2006 2010s

Decade Decade Year Year Decade

Time Sharing Virtual private Cloud metaphor for Commercialisation of IoT, Fog and other
network (VPN) virtualised services Cloud computing
paradigms
 INTRODUCTION TO CLOUD COMPUTING
Revisiting Cloud Computing

Ubiquitous, Convenient,
On-demand
Servers
Minimal managing efforts

Network
Application
Storage
Services

Rapidly provisioned
INTRODUCTION TO CLOUD COMPUTING
NIST Definition

“Cloud computing is a model for enabling ubiquitous,

convenient, on-demand network access to a shared
pool of configurable computing resources (e.g.,
networks, servers, storage, applications, and services)
that can be rapidly provisioned and released with
minimal management effort or service provider
interaction.”
INTRODUCTION TO CLOUD COMPUTING
Cloud Categories

Cloud Features and Categories

Characteristics:
● Measured service
● On-demand self-service
● Broad network access
● Resource pooling
● Rapid elasticity
Service Models:
● IaaS, PaaS, SaaS
Deployment Models:
● Public, Private, Community and Hybrid
Cloud Computing Architecture
 CLOUD COMPUTING ARCHITECTURE
Overall Architecture

Client Front end

Internet

Application

Management Service Security Back end

Storage
 CLOUD COMPUTING ARCHITECTURE
Client

Client Front end

Client:
Internet
End-user of the
application
Provides interface to
access the application
Application
Includes applications
required to access cloud

Management Service Security Back end

Storage
 CLOUD COMPUTING ARCHITECTURE
Application

Client Front end Application Layer:

The application layer
consists of applications
that are deployed on the
Internet
cloud.
Types of Applications:
● Games
Application ● Platform-based: Web
and mobile
● Technology-based:
Management Service Security Back end IoT and machine
learning

Storage
 CLOUD COMPUTING ARCHITECTURE
Service

Service Layer:
Client Front end
Handles services
required for installing,
configuring and
Internet managing a basic
application.
Example:
● Networking
Application
● Instances
● Operating system
Management Service Security Back end ● Storage
● Security

Storage
 CLOUD COMPUTING ARCHITECTURE
Storage

Storage Layer
Client Front end Functionality
● It stores any type of
unlimited data in the
Internet cloud environment
from different
sources, such as
company data or web
portals.
Application
Characteristics:
● Secure storage

Management Service Security Back end ● On-demand access

● Versioning enabled

Storage
 CLOUD COMPUTING ARCHITECTURE
Management

Client Front end

Management Layer
Responsibility:

Internet System security

● Access
● Authentication
● Authorisation
Application
Storage
Application

Management Service Security Back end Memory

Storage
 CLOUD COMPUTING ARCHITECTURE
Security

Client Front end Security Layer

Functionality:
● It provides high-level
Internet security to clients at
different locations
and protects data
from intruders and
disasters.
Application
Characteristics:
● Algorithms include
AES 256-bit key and
Management Service Security Back end RSA.

Storage
INTRODUCTION TO CLOUD COMPUTING
Cloud Service Providers - Examples

Amazon Web Services (AWS)

IBM Bluemix
Microsoft Azure
Yahoo Pipes
Google App Engine
Salesforce
Bitrix 24
 APPLICATIONS OF CLOUD COMPUTING
Real-World Applications

YouTube
Microsoft OneDrive
Dropbox
Slack
Gmail
Google Drive
Cloud Service Models
 CLOUD SERVICE MODELS
Traditional IT On-Premise Solution

Traditional IT
Application

Data
Traditional IT services requires you to manage
Runtime everything related to your application.
You manage

Middleware ● Example: Schools9.com employs traditional

Operating system IT services instead of Cloud.

Virtualization
● The services shown in the diagram on the left
must have to be managed by the admin or
Servers the developer of the application.
Storage

Networking
 CLOUD SERVICE MODELS
Types

Cloud Service
Models

Infrastructure-as-a-Service Platform-as-a-Service Software-as-a-Service

(IaaS) (PaaS) (SaaS)
 CLOUD SERVICE MODELS
Infrastructure-as-a-Service (IaaS)

Infrastructure
(as a service) It provides high-level APIs to implement low-level applications.
It provisions hardware/infrastructure requirements to customers.
Application
It enables platform virtualisation for users to support software
You manage

Data development.

Runtime Services offered by IaaS:

● Physical computational resources
Middleware
● Location
Delivered as a service

Operating system ● Data partitioning

Virtualization ● Scaling

Servers ● Security
● Backup
Storage
Examples: Amazon Web Services (AWS), Microsoft Azure, Google
Networking Compute Platform (GCP)
 CLOUD SERVICE MODELS
Platform-as-a-Service (PaaS)

Platform
(as a service)
manage

Application PaaS provides a cloud-based environment to:

You

Data ● Run the code,

Runtime ● Build and deliver the application,
● Manage and develop the application.
Delivered as a service

Middleware

Operating system The service provider supplies the platform and

infrastructure.
Virtualization
The user can maintain the software.
Servers
Examples: OpenShift, Elastic BeanStalk, Google App
Storage Engine
Networking
 CLOUD SERVICE MODELS
Software-as-a-Service (SaaS)

Software
(as a service)

Application

Data
The application, security, access, etc. are all
Delivered as a service

Runtime
managed by the cloud service provider.
Middleware The users do not need to install any
Operating system programs/applications.
Virtualization Examples: Gmail, Dropbox, Microsoft OneDrive,
Cisco Webex, GoToMeeting
Servers

Storage

Networking
 CLOUD SERVICE MODELS
IaaS vs PaaS vs SaaS

Infrastructure Platform Software

Traditional IT (as a service) (as a service) (as a service)

You manage
Application Application Application Application

You manage
Data Data Data Data

Delivered as a service
Runtime Runtime Runtime Runtime
You manage

Delivered as a service
Middleware Delivered as a service Middleware Middleware Middleware

Operating system Operating system Operating system Operating system

Virtualization Virtualization Virtualization Virtualization

Servers Servers Servers Servers

Storage Storage Storage Storage

Networking Networking Networking Networking

 CLOUD SERVICE MODELS
Specialised Service Models

Cloud Services Delivery Models

Examples of such models include:
● Data-as-a-Service (DaaS) - Variant of PaaS
● Database-as-a-Service (DBaaS) - Variant of PaaS
● Analytics-as-a-Service (AaaS) - Variant of SaaS
● Communication-as-a-Service (CaaS) - Variant of IaaS
● Anything-as-a-Service (XaaS) - It can be a variant of IaaS, SaaS or PaaS.
Cloud Deployment Models
 WHAT IS A CLOUD DEPLOYMENT MODEL?

It is a specific configuration of environmental parameters, such as:

● Accessibility,
● Proprietorship of infrastructure, and
● Storage size
To capitalise on this computing type, choose a model that best suits your requirement.
To choose an appropriate deployment model, consider the following:
● Technical requirements: Storage, networking and computing;
● Business goals;
● Available resources; and
● Appropriateness of use-case requirements.
 CLOUD DEPLOYMENT MODELS

Cloud Deployment
Models

Public Cloud Private Cloud Community Cloud Hybrid Cloud

 PUBLIC CLOUD

Accessibility: General public

Storage: Third-party servers
Advantages:
● High scalability
● High availability
● Easy management
● Reduced costs
Disadvantages: Cannot be customised
 PRIVATE CLOUD

Accessibility: Companies
Storage: Dedicated infrastructure
Advantages:
● High privacy and security
● High reliability
● Bespoke (customisable)
Disadvantages: Expensive (high costs)
 COMMUNITY CLOUD

Accessibility: Group of organisations

Storage: Infrastructure shared
between the owners
Advantages:
● Ease of data sharing
● Costs lower than private cloud
● Better privacy, security and
reliability than public cloud
Disadvantages:
● Costs higher than public cloud
● Privacy, security and reliability not as
good as private cloud
 HYBRID CLOUD

It refers to mixed computing. It combines

the features of any other deployment
models.
Community cloud is an example of
Hybrid Cloud.
Example: A company can use a
deployment model, wherein it uses a
secure private cloud for confidential
processes/workloads and a public cloud
for everything else.
Fog Computing
 WHAT IS FOG COMPUTING?

A globally disparate computing architecture that ubiquitously connects

numerous heterogeneous devices at the edge of the network to collaboratively
provide flexible computing, communication and storage services.
● It is also called fog networking or fogging.
● It is a decentralised computing infrastructure in which the computing
resources are located between the cloud and the data source.
● It is used for data computing, storage and applications.
● It extends cloud computing to the network's edge.
 WHY FOG COMPUTING?

Pooling of local resources

Reducing the amount of data transmissions
Performing real-time data analytics and reducing latency
Easing bottlenecks in the network bandwidth
Enabling new class of applications and services
FOG COMPUTING CHARACTERISTICS

Low latency
Location awareness
Inclusion of a considerable number of devices
Widespread topographical dispersal
Heterogeneity
Mobility
 FOG COMPUTING ARCHITECTURE

The hierarchical fog architecture consists of three layers:

IoT (Terminal) Layer:
● It is the nearest layer to the end-user.
● It consists of IoT devices (sensors, smart cars, cell phones, readers, smart cards, etc.).
Fog Layer:
● This layer consists of multiple fog nodes.
● It includes routers, base stations, gateways, switches, fog servers, access points, etc.
● Fog nodes and cloud data centre connections are provided by the IP core networks.
Cloud Layer:
● It comprises of high-performance cloud servers and storage devices.
● It offers real-time application services (e.g., smart transport, smart home, smart factory, etc.) to the users.
 FOG COMPUTING HIERARCHICAL ARCHITECTURE

Cloud layer Core

Cloud gateway

Fog layer

Access Point

IoT layer

Edge
 EXAMPLES OF FOG COMPUTING APPLICATIONS

A machine-to-machine (M2M) communication or human-machine

interaction (HMI) can be initiated by fog computing by the monitoring
and analysing of real-time data from a network of connected things.
A few examples include:
● Locking a door
● Changing equipment settings
● Applying brakes on a train
● Zooming a video camera
● Opening a valve in response to a pressure reading
● Creating a bar chart
● Alerting a technician for a preventive repair
 FOG VS CLOUD
Parameter Fog Cloud

Goal Enhance proficiency and execution of process that Give a request of greatness change in the
should be transported to the cloud for handling, practical, powerful provisioning of IT
investigation and storage administrations

Architecture Distributed Centralised

Communication with devices Directly from the edge From a distance

Response time Seconds to minutes Minutes, days, weeks

Application examples Visualisation and simple analytics Big data analytics, Graphical dashboards

How long data is stored? Short duration: perhaps hours, days, or weeks Months or years

Geographic coverage Wider compared to IoT Global

Support of operating system (OS) Hypervisor virtualisation A hypervisor (VMM) on which multiple OSs run

Latency + Security Low/high High/low

Connectivity Various protocols and standards Internet protocol (IP)

Interoperability Interoperability between heterogeneous resources Web services (SOAP and REST)
 Cloud Support for Blockchain,
Artificial Intelligence, Analytics,
Software Development
 CLOUD SUPPORT FOR BLOCKCHAIN

Blockchain is a collection of distributed ledgers for recording and tracking the value of
a commodity.
When a new data is added, a block with a proof of work (PoW) is created.
● PoW is a hash value.
● Ledgers/transactions are unchangeable, unless PoWs of the preceding blocks
are changed.
Miners: They mine blocks, generate and validate PoWs.
● PoW is completed.
● New block is added to the network.
● Rest of the network validity is verified and added to the blockchain.
Example: Cloud domain and other real-time data security systems for ensured
data protection.
‘TRACK AND TRACE’ USE CASE OF BLOCKCHAIN ON AWS/CLOUD

Track and Trace:

Identifies the past/present locations of the product inventory
Traces the product’s origin and tracks it from the source to the consumer
Lack of data compatibility in supply chain leads to:
● Visibility gaps
● Manual errors
● Inaccurate supply and demand predictions
● Compliance violations
● Counterfeiting
Amazon/Cloud-managed blockchain enables the documentation of
network-updates to single shared ledger.
Advantages: Total data visibility, single source of truth
Example includes Product authenticity verification
 ‘TRACK AND TRACE’ USE CASE OF BLOCKCHAIN ON AWS/CLOUD

Challenges Benefits

Supplier: Ethical sourcing uncertain Supplier: Ethical sourcing ensured

Manufacturer: Environmental impacts Manufacturer: Measure and reduce

unknown. Fragmented data systems and environmental impacts. Complete data
data loss visibility on single shared ledger

Regulator: Reduced need for regulators.

Regulator: Infrequent third party quality
Smart contracts ensure quality control
control
throughout the supply chain

Logistics: Manual transport updates not Logistics: Automated real time transport
in real-time updates
 ‘TRACK AND TRACE’ USE CASE OF BLOCKCHAIN ON AWS/CLOUD

Challenges Benefits

Wholesaler: Overstock and stockouts

Wholesaler: Manage stock with real-time
due to inaccurate supply and demand
data
data

Retailer: Product provenance and Retailer: Certainty of product

authenticity uncertain provenance and authenticity

Consumer: Supply chain insight

Consumer: Minimal supply chain insight
empowers informed decision making
 CLOUD SUPPORT FOR ARTIFICIAL INTELLIGENCE

Network objectives for the deployment of artificial intelligence (AI) on the

IoT/fog/cloud nodes:
● Workload and environment awareness
● Continuous adaptation for better quality of service (QoS)
● Power consumption reduction
● Infrastructure cost reduction
AI comprises of the following:
● Search algorithms
● Machine learning
● Reinforcement learning
● Planning
AI performs data-intensive tasks, such as:
● Optimal task scheduling decisions
● Optimisation of virtual machine (VM) migrations under constraints, such as
computation capabilities, bandwidth limits for SLA, and task deadline requirements
CLOUD SUPPORT FOR ARTIFICIAL INTELLIGENCE

AWS/Cloud Pre-Trained Machine Learning Models

They address common use case scenarios, such as:
● Personalised recommendations
● Modernised contact centres
● Enhanced safety and security
● Increased customer engagement
Example: Yotascale provides the following solutions:
● Automation of performance management, thereby reducing
the accountability on humans
● Predictions for the future by examining historical data
● Manage and create cloud cost visibility
● Optimisation of cloud costs
● Analysis of data in real time
COMMON CLOUD SERVICES FOR ARTIFICIAL INTELLIGENCE

Advanced text analysis (Amazon Comprehend)

Automated code reviews (Amazon CodeGuru)
Chatbots (Amazon Lex)
Demand forecasting (Amazon Forecast)
Document analysis (Amazon Textract)
Personalized recommendations (Amazon Personalize)
Image and video analysis (Amazon Rekognition)
 CLOUD SUPPORT FOR ANALYTICS

Examines datasets and draws conclusion

Data analytics techniques take raw data and uncover patterns to
extract valuable insights from it.
Here are two of the broader categories:
Data Generation:
● Velocity
● Scalability
● Dynamics
● Heterogeneity
Data Quality:
● Uncertainty
● Redundancy
● Ambiguity
● Inconsistency
COMMON CLOUD SERVICES FOR ANALYTICS

Interactive analytics (Amazon Athena)

Big data processing (Amazon EMR)
Data warehousing (Amazon Redshift)
Real-time analytics (Amazon Kinesis)
Operational analytics (Amazon Elasticsearch Service)
Dashboards and visualizations (Amazon QuickSight)
Visual data preparation (AWS Glue DataBrew)
 CLOUD SUPPORT FOR SOFTWARE DEVELOPMENT

Cloud Support for Software Development: It includes host code,

build, test and deploy applications to the cloud

Use Cases:
Continuous Integration:
● It is a DevOps software development practice where code
changes are merged into a central repository, after which
automated builds and tests are run.
Continuous Delivery:
● It is a DevOps software development practice where code
changes are automatically prepped. After the build stage, the
code changes are deployed to a testing and production unit.
 CLOUD SUPPORT FOR SOFTWARE DEVELOPMENT

Use Cases:
Infrastructure as Code:
● Models the applications and infrastructure as code,
● Automates the provisioning of infrastructure services, and
● Eliminates the need for manual updates for error-prone
infrastructure.
Monitoring and Logging:
● Monitors metrics and logs by observing the applications and
infrastructure performance for better experience.
● Captures, categorises and analyses the data to understand the
impact of code changes.
COMMON CLOUD SERVICES FOR SOFTWARE DEVELOPMENT

Amazon Corretto
AWS Cloud9
AWS CodeBuild
AWS CodeArtifact
AWS CodeDeploy
AWS CodePipeline
AWS CodeStar
ACTIVITY: DEVELOPING AN APPLICATION ON AWS DYNAMODB

Login to AWS and select AWS DynamoDB

01 Database Development Service.

On the AWS DynamoDB page, choose Create

02 table.

On the Create DynamoDB table template page,

03 enter table name and primary key.

Trigger a CloudWatch alarm for this table under

04 the Triggers option.
DEMO: DEVELOPING AN APPLICATION ON AWS DYNAMODB

Choose the Create Trigger using New function or

05 existing lambda function.

06 Ensure that the lambda function is accessible.