Lecture 1 : Introducing Cloud and Edge

Computing

Dr. Bibhas Ghoshal

Assistant Professor

Department of Information Technology

Indian Institute of Information Technology Allahabad

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Resources
Course website : [Link]
●
Slides, Important links, research articles, homework and lab assignments will be provided here
Recommended Books
1. Distributed and Cloud Computing : From Parallel processing to Internet of Things
Kai Hwang, Jack Dongarra, [Link]
Morgan Kauffman, 2011

2. Cloud Computing : Principles and Paradigms

Edited by Raj Kumar Buyya, James Broberg, Andrzej Goscinski
Wiley, 2011
3. Research Papers
(will be provided on the course website time to time)
●
Slides provide key concepts, books provide details
●
Use lecture notes supplemented with textbook and internet

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Resources for Tut and Lab

Software Environments

Web Services and languages

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 What is the course all about ???
●
Cloud Computing Technology
● Cloud computing basics – overview, architecture, deployment
● Virtualization
● How to use cloud tools, APIs, SDKs
● How to access cloud – cloud simulator
●
Cloud Computing as Distributed Systems Environment
● How to design Applications for cloud
● Web services, Service oriented architecture
● XML, SOAP, WSDL and REST
● Managing Data, Economics
● Open Stack
● Map Reduce
● Case study with commercial clouds - Microsoft Azure and Google Cloud Platform
●
Cloud Computing as a Research Topic
● Cloud Security, Federated
● Fog Computing
● DOCKER Container
● Cloud Computing Challenges
●
Edge Computing
● Edge Computing Basics – Architectures, Applications, Edge Intelligence
● IoT Cloud
● Mobile Computing

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
What is a computer ?

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 History of Computing

Babbage Difference Engine, 1879 IBM 360 Mainframe, 1964

Harvard Mark-I, 1944 MIT Whirlwind, 1951
$2.5-3 million

DEC PDP-8 minicomputer, 1965,

$18,000 ATARI micro computers Gaming + home computing, 1979

Slide Adapted from : Virtualization andCloud Computing, Sourav Bansal, [Link]

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
History of Computing

Apple Machintosh, 1984 IBM PC/AT, 1984 Microsoft Windows, 1985

Intel 80286 Microprocessor

Internet and Search,

1998

Linux, 1991 IPAD, 2010

Intel Pentium Processor Diagram,1993
Slide Adapted from : Virtualization andCloud Computing, Sourav Bansal, [Link]

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Trends in Computing

1. Distributed Computing
2. Grid Computing
3. Cluster Computing
4. Utility Computing
5. Cloud Computing

33

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Distributed Computing
Distributed Computing is a method in which different parts of a computer
program run simultaneously on two or more computers that are
communicating with each other over a network.

Traditional Computing Components:

Data Storage Core Logic User Interface
1. Database 1. Application Specific 1. GUI
[Link] based 2. Algorithm Implementation 2. Command Line
All Components sit on the same computer

Distributed Computing : Core Logic

1. Application Specific
2. Algorithm Implementation
Data Storage
1. Database User Interface
[Link] based33 1. GUI
2. Command Line
Cloud and Edge Computing
Spring 2023
Instructor : Dr. Bibhas Ghoshal
Characteristics of Distributed Computing

1. Each node plays a partial role

2. Fault Tolerance

3. Transparency
4. Resource sharing

5. Load Sharing

6. Easy to expand

7. Performance 33

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Cluster Computing
Cluster Computing : Type of parallel or distributed computer systems which consists of stand-alone computers
working together as a single integrated computing resource

Clusters are deployed to improve speed and/or reliability over that provided by a single computer while
typically being much more cost effective than single computer of comparable speed or reliability

Characteristics :

i. Network is faster than typical LAN

ii. Low latency Communication protocols
iii. Loosely coupled than Shared Multiprocessors

Types of Cluster :

i. High availability or Fail over cluster

ii. Load Balancing cluster
iii. Parallel/Distributed Processing cluster

33

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Big Computers
Sun E-10000 “supermini”
i. Up to 64 processors @250MHz
ii. Up to 64 GB RAM
iii. Up to 20 TB Disk
iv. Used by eBay, among others

PC
200 MHz CPU, 32MB RAM, 4 GB disk

UC Berkeley Networks Of Workstations (1994-1999)

NOW2,
NOW0, NOW1, 1997
1994 1995 60 SUN
4 HP-735’s 32 SUN Spark-2
Spark
33
Stations
Cloud and Edge Computing
Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Grid Computing

Grid Computing : combination of computer resources from multiple administrative domains

applied to a common task *

Core idea
-Tdistributed
heGrid :
parallel computation
- super virtual computer
●
Electrical Grid Analogy
Electrical Power Grid Grid
Users get access to Users get access to
electricity through wall computing resources
sockets with no care of ( processor, storage, data
where or how the electricity applications) as needed
is generated with no knowledge of
where the resources are
located

The power grid links power The grid links computing

plants of different kinds resources and provides
mechanism to access them
33
* Ian Foster's and Carl Kesselman's seminal work, "The Grid: Blueprint for a new computing infrastructure" (2004)
Cloud and Edge Computing
Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Grid Computing
Characteristics:
1. Share more than information - data, computing power, applications in dynamic environment, multi-institutionaland
virtual organization
2. Efficient use of resources – People from many institutions work to solve a problem
3. Join local communities
4. Interactions with layer underneath must be transparent

Need for Grid Computing :

1. Exploiting under utilized resources
2. Data Visualization is important
3. Scientific problems are becoming complex and users need more accurate and precise solutions to their problems in
minimum time – need scientific collaboration

Uses of Grid Computing :

1. Weather forecast
2. Detection and Modelling of natural disasters
3. Physics applications
4. Crystallography

Type of Grids :
1. Computational Grid
2. Data Grid
3. Collaboration grid 33

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Data Centre Computing
When computing needs exceed computing resources, there are 2 options (not mutually
exclusive)

1. Scale-up (vertical scaling): Increase resources of individual server

2. Scale-out (horizontal scaling): Increase the number of servers

Modern Data Centres lean towards Scale-out

Data Center Architecture

33

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Data Centre

2005: Datacenter is new “ server”

“ Program” => Web search, email, map/GIS, …
“ Computer” => 1000’s computers, storage, network
Warehouse-sized facilities and workloads

Figures : Sun Microsystems, CNET, & [Link]

33

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Data Centre as a Computer?

Facebook Datacenter, Clonee, Ireland Google Datacenter, USc

Inside the Google Datacenter

Figure Source : Internet

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Utility Computing Arrives
- Amazon Elastic Compute Cloud (EC2)
- Compute unit rental: $0.08-0.80/hr.
1 CU ≈ 1.0-1.2 GHz 2007 AMD Opteron/Xeon core

Instances Platform Cores Memory Disk

Small - 32-bit 1 1.7GB 160

$0.10/hr GB

Large - 64-bit 4 7.5GB 850

$0.40/hr GB

XLarge - 64-bit 8 15.0GB 1690

$0.80/hr GB

Billing rounded to nearest hour; pay-as-you-go storage also available

A new paradigm for deploying services
“ Computing as Utility” – A vision from MULTICS c. 1969

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Computing Today

SaaSComputing

The Million Server Datacenter

Providing IaaS
Slide Adapted from : Virtualization andCloud Computing, Sourav Bansal, [Link]

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
What exactly is Cloud Computing?
“ It is Nothing New “

“ ...we’ ve redefined Cloud Computing to

include everything that we already do...
I don’t understand what we would do
differently ... other than change the
wording of some of our ads.”

– Larry Ellison, CEO, Oracle

(Wall Street Journal, Sept. 26, 2008)

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Must Read !!!

Above the Clouds:

A Berkeley View of Cloud
Computing
(February 10, 2009)

[Link]

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Cloud Computing
●
Old Idea : Software as a Service (SaaS)
Software hosted on local servers or desktops (dumb terminals)
●
New : pay-as-you go utility computing
Illusion of infinite resources on demand
Fine grained billing : release == don’t pay; Services – longer commitment, more $$$/hour
Public vs. Private cloud

●
Why Now?
i. The Web “ Space Race” : Build-out of extremely large datacenters (10,000’s of commodity PCs)
ii. Driven by growth in demand (more users) -
Infrastructure software : e.g., Google File System
Operational expertise
Discovered economy of scale: 5-7x cheaper than provisioning a medium-sized (100’s machines) facility
iii. More pervasive broadband internet
iv. Open source software

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Cloud Computing - NIST
Cloud Computing is a model for enabling convenient, on
demand network access to a pool of configurable
computing resources ( eg. Network, server, storage,
applications, and services) that can be rapidly provisioned
and released with minimal management effort or service
provider interaction. This cloud model promotes
availability and is composed of five essential
characteristics, three service models, and four deployment
models

[Link]

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Cloud Economics
●
Static Provisioning for peak – wasteful but necessary

Slide Adapted from : Above the Clouds:A Berkeley View of CloudComputing (February 10, 2009)

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Risk of UnderUtilization
●
Underutilization results if peak predictions are too optimistic

Slide Adapted from : Above the Clouds:A Berkeley View of CloudComputing (February 10, 2009)

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Risk of Under Provisioning

Slide Adapted from : Above the Clouds:A Berkeley View of CloudComputing (February 10, 2009)

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Cloud Computing Vs. Grid Computing
●
Distinctions not clear because Cloud and grid share similar visions
- Reducing computing costs
- Increasing flexibility and reliability by using third-party operated hardware
●
Resource Sharing
- Grid : collaboration
- Cloud : assigned resources not shared
●
Virtualization
- Grid : Virtualization of data and computing resources
- Cloud : virtualization of hardware and software platforms
●
Security
- Grid : security through credential delegation
- Cloud : security through isolation
●
Self Management
- Grid : reconfigurability
- Cloud : reconfigurability and self healing
●
Payment mode
- Grid : rigid
- Cloud : flexible

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Cloud Computing Characteristics
●
Cost Associativity
- 1,000 CPUs for 1 hour same price as 1 CPU for 1,000 hours
●
On-demand self service
Consumer can unilaterally provision computing capabilities without human interaction with service provider
●
Broad network access
Capabilities are available through network and suing standard mechanisms
●
Multi-tenancy / Resource pooling
Provider’s resources are pooled to serve multiple customers
●
Rapid elasticity
capabilities can be rapidly and elastically provisioned
Unlimited virtual resources
Predicting a ceiling is difficult
●
Measured service

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Relevant Technologies in Cloud Computing
●
Access
- heterogenous set of thick and thin clients
- high speed broad band access
- Data centres ( large computing capacity )
●
Virtualization ( Decoupling from physical computing resources)
- Types : Hardware , Memory, Storage, Network
- Hardware : emulation (QEMU) , para-virtualization (Xen), full virtualization(VMware)
- Memory : Decouples RAM resources from individual systems
Aggregates these resources into a virtualized memory pool available to any computer
- Storage : abstracting logical storage from physical storage
- Data Virtualization : Data as an abstract system independent of underlying data base systems, data structure
And storage
- Network : Virtualized Network addressing space with or within or across network subnets; VPN
Q : How do you measure Virtual Resources
Amazon ECU ( Elastic Compute Unit) : EC2 Compute unit equals 1.0 – 1.2 GHz 2007 Opteron or 2007 Xeon processor
●
APIs
Required for various operations and applications; No standards

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Cloud Service Taxonomy
●
Layer
Hardware – as - a Service (Haas)
Infrastructure -as -a Service (IaaS)
Data Storage – as- a- Service (DaaS)
Communication-as-a Service (CaaS)
Platform-as-a Service (PaaS)
Software-as-a Service (SaaS)

●
Type
Public Cloud
Private Cloud
Hybrid Cloud

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Software as a Service ( SaaS )
●
Vendor / Provider controlled application
accessed over the network

●
Characteristics :
- Network based access
- Multi-tenancy
- Single software release for all
- browser based client and client tools

●
Disadvantages :
- dependency on network and cloud provider
- limited client bandwidth
- cloud service provider in charge of the data

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Platform as a Service ( PaaS )
●
Vendor provided deployment environment
tools and technology selected by the vendor
( control over data life cycle )

●
Characteristics :
- Rapid development and deployment
- Small startup cost
- multi-tenancy
- load balancing and fail over
- native integrated environment – performance, load

●
Disadvantages :
- inherits all disadvantages from SaaS
- choice of technology limited to vendor provided tools and services

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Infrastructure as a Service ( IaaS )
●
Vendor provided and consumer provisioned computing resources
●
Characteristics :
- Consumer is provided customized virtual machines
- Consumer has control over OS, memory, storage, servers and
deployment configurations
- limited control over network resources
- infrastructure scalability
- economical cost

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Cloud Services and Control

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Cloud Services Reference Model

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Cloud Deployment Models
●
Private Cloud
- infrastructure operated by an organization
- managed by organization and third party

●
Community cloud
- supports a specific community
- infrastructure shared by several organizations

●
Public cloud
- infrastructure is made available to public
- owned by an organization selling cloud services

●
Hybrid cloud
- infrastructure is a composition of two or more cloud deployment models
- enables data and application portability

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Cloud Distributed Storage
●
Distributed Storage : Two approaches of scaling
- Vertical - bigger hardware
- Horizontal – more hardware
( horizontal partitioning, functional partitioning)
●
CAP Theorem *
- web services cannot ensure all the three of the following properties
at once :
- consistency : set of operations has occurred all at once
- availability : an operation must terminate in an intended response
- partition tolerance : operations will complete even if individual components are unavailable

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Cloud Distributed Storage
●
General Characteristics :
- Simplified data model
- Built on Distributed File System
GFS – Google File System
HDFS – Hadoop Distributed File System

●
Highly Available – Relaxed Consistency
●
Fault-Tolerant – Replication
●
Eventual Consistency – all replicas will be updated in different time and in different order

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Cloud Distributed Computation
●
Motivation
- distributed computing – thousands of computers
- large data sets
- fault tolerant
- easy to configure and manage
●
Basic idea : functional programming
- large problem broken into simpler problems ( functional decomposition )
- each small problem can be solved by a functional transformation of input
data
- each problem can be executed in complete isolation
●
Server to solve big problem

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
 Cloud Distributed Computation
Distributed grep

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Cloud Distributed Computation
Distributed word count (wc)

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Cloud Distributed Computation : Map Reduce

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Obstacles and Opportunities for Cloud
Computing

Slide Adapted from : Above the Clouds:A Berkeley View of CloudComputing (February 10, 2009)

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Cloud Computing for IoT
Cloud is an IoT enabler :
Huge amount of IoT Data needs storage, retrieval and management ( sustained services)
Fast Analytics for Prediction and Critical Decision making

Benefits of Cloud in IoT :

Scalability ; Data Mobility; Time to Market; Security; Cost effectiveness

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Challenges in Using Cloud Platform for IoT

Source : Internet

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Fog Computing

Source : Internet

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Edge Computing

Benefits of Edge Computing :

1. Faster Response Time

2. Reliable Operations with

intermittent connectivity

3. Security and Compliance

4. Cost-Effective Solutions

5. Interoperability between
Legacy and Modern
Devices

Source : Internet

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Edge AI

Edge Computing + Artificial Intelligence = Edge AI

AI algorithms are executed locally on a hardware device using the data collected from Edge computing.

AI computation is done near the user at the edge of the network, close to where the data is located

Benefits :

1. Reduces latency
2. Reduces power consumption
3. Reduces Data Cost
Source : Internet

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Edge AI Vs. Cloud AI

Source : Internet

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Edge AI in Use

Source : Internet

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal
Researh Challenges in Edge AI

Source : Edge Intelligence: Architectures, Challenges, and Applications; Dianlei Xu et al.

Cloud and Edge Computing
Spring 2023
Instructor : Dr. Bibhas Ghoshal
Register yourself on Google Classroom
For any queries : [Link]@[Link]
Web : [Link]/[Link]

Cloud and Edge Computing

Spring 2023
Instructor : Dr. Bibhas Ghoshal