Introduction to Cloud

Computing
(Module 1)
Dr. Hemangi Goswami
Assistant Professor, School of Computer Engineering
Manipal Institute of Technology Bengaluru
Manipal Academy of Higher Education, Manipal, INDIA
E-mail: [Link]@[Link]
Outline

 Cloud computing in a Nutshell

 Roots of Cloud Computing
 Cloud computing, its inception, and Desired
Features
 Cloud Architecture
 Cloud Service Models
 Cloud Deployment models
 Challenges and risks

2
Cloud computing in a Nutshell

3
Cont.
•Cloud Computing as the Fifth Utility:
•Revolutionized the computing paradigm.
•Recognized as the fifth utility after water, electricity, gas, and telephony.
•Comparison with Cluster and Grid Computing:
•Unlike Cluster and Grid computing, Cloud is an amalgamation of Parallel and
Distributed computing.
•Tasks are broken down and shared among heterogeneous resources/processors.
•Enables concurrent scheduling of tasks.
•Cloud as a Milestone in Computational Advancements:
•Focuses on:
•Designing computing systems.
•Developing scalable applications.
4
•Provisioning existing resources on demand.
Core Principles of Cloud Computing
[Link] Provisioning:
•Automatically expands or contracts resource capacity based on real-time demand.
[Link]:
•Provides:
•Resource isolation.
•Emulation of hardware/software.
•Customization and security.
•Enables on-demand delivery of computing services.
•Economic and Service-Oriented Nature:
•Functions as a milestone in the global economy.
•Offers membership-based services, accessible anytime and anywhere.
•Pricing Model:
•Follows a "pay-per-use" strategy.

5
 Roadmap of
Computing
Paradigm

6
Fig. 1: Computing paradigm evolution
 Parallel
Computing
 Type of computation in
which many calculations
or processes are carried
out simultaneously.
 Divides a large problem
into smaller sub-
problems, solving them
concurrently using
multiple processing
units (such as CPUs,
GPUs, or a cluster of
computers), and
combining the results to
Fig. 2: Parallel Computing produce the final output.
[1]
7
8
Parallel Computing (contd.)
Features:
 Concurrency: Multipletasks or processes are
executed at the same time.
 Task Division: A problem is divided into smaller
independent or interdependent tasks.
 Multiple Processing Units: Tasks are distributed
among different processors or cores.
 Faster Computation: Parallelism reduces
execution time for large-scale computations.
9
Parallel Computing (contd.)
Benefits:
 Improved Performance: Reduces time required for
complex computations.
 Scalability: Can handle larger datasets by adding
more processing units.
 Efficiency:Makes better use of available
resources, especially in multi-core and
distributed systems.

10
Cluster Computing
 A cluster is a type of parallel or distributed computer
system, which consists of a collection of inter-
connected stand-alone computers working together
as a single integrated computing resource.

 Clusters are usually deployed to improve speed

and/or reliability over that provided by a single
computer, while typically being much more cost
effective than single computer.
11
 Cluster
Computing
(contd.)
Types of Cluster:
 High Availability or Failover
Clusters
 Load Balancing Cluster
 Parallel/Distributed
Processing Clusters

Cluster Components:
 Basic building blocks of
clusters are broken down
into multiple categories:
 Cluster Nodes

Fig. 3: Cluster  Cluster Network

12
Computing [2]  Network Characterization
Cluster Computing (contd.)
Benefits:
 Improved Performance: Tasks are processed faster
by distributing workloads across multiple
nodes.
 Fault Tolerance: Failure of one node does not
disrupt the entire system.
 Flexibility: Clusters can be tailored for specific
applications or workloads. (scale modify upgrade
clusters)
13
Cluster Computing (contd.)
Limitations:
 High initial setup cost
 Complex management
 Scalability issues: While clusters are scalable, adding
more nodes can lead to diminishing returns due to
communication overhead and data synchronization
delays.
 Network dependency
 Limitedfault tolerant: Although clusters provide
redundancy, a significant hardware failure (e.g.: a
switch or network failure) can disrupt the entire 14
Distributed Computing
 A computing paradigm where a single system is
composed of multiple autonomous computers or
nodes that communicate and coordinate their actions
by passing messages to achieve a common goal.

 These systems work together to process tasks, share

resources, and solve problems that a single computer
cannot handle efficiently.

15
Distributed Computing
(contd.)
Features:
 Multiple Nodes: Tasks are distributed across a
network of independent computers.
 Decentralization: There is no single centralized
control; instead, the system relies on
cooperation among nodes.
 Scalability: Systems can scale horizontally by
adding more nodes to handle increasing
workloads.
 Fault Tolerance: The failure of one or more nodes 16
Distributed Computing
(contd.)
Benefits:
 Resource Sharing: Efficient utilization of
resources, including processors, storage, and
network bandwidth.
 Scalability: Accommodates large datasets and
computations by adding more nodes.
 High Availability: Redundancy in nodes ensures
that the system remains operational even if
some nodes fail.
 Reduced Latency: Processing tasks closer to 17
 Distributed
Computing
vs.
Centralized
Computing
 Early computing were
performed on a single
processor
 Uni-processor
computing can be
called as centralized
computing.
 Tightly coupled
(Centralized
Computing)
 Loosely coupled
(Distributed Computing)

Fig. 4: Centralized vs. Distributed 18

Computing
 Grid
Computing
 It involves pooling
together the resources
(resource sharing) of
multiple geographically
distributed and
heterogeneous
computers to work
collaboratively on large-
scale tasks.
 Unlike cluster
computing, where the
nodes are tightly
coupled and located
within the same physical
Fig. 5: Working of Grid space, grid computing
19
computing [4] systems are loosely
Grid Computing (contd.)
Features:
 Resource Sharing: Combines computational power,
storage, and network resources from different systems,
often across organizations or institutions.
 Heterogeneity: Resources can have different operating
systems, hardware configurations, and software
environments.
 Geographic Distribution: Resources are spread across
multiple locations, connected via the internet or wide-
area networks (WANs).
 Scalability: New resources can be added to the grid with
minimal reconfiguration.
 Autonomy: Each participating system maintains its own
administrative control.
20
Grid Computing (contd.)
Benefits:
 Enhanced Computational Power: Combines
distributed resources to solve large-scale,
complex problems.
 Cost Efficiency: Utilizes existing infrastructure,
reducing the need for dedicated high-
performance computing systems.
 Flexibility: Can dynamically allocate resources
based on the task requirements.
 High Availability: Provides redundancy and fault
21
tolerance, ensuring tasks continue even if some
Grid Computing (contd.)
Limitations:
 Complexity: Managing, scheduling, and securing a
large, distributed grid is challenging.
 Network Dependency: Relies
on consistent and
high-speed internet; network delays can affect
performance.
 Security: Shared resources across organizations
can lead to data and privacy concerns.
 Resource Contention: Shared resources might lead
to conflicts between users or tasks.
 Software Compatibility: Applications need to be
22
specifically designed or adapted for grid
Table: Cluster Computing vs. Grid Computing vs. Cloud
Computing

Cluster
Computing
vs.
Grid Computing
vs.
Cloud Computing

23
The Next Revolution in IT

24
 Cloud Computing

 A model for delivering computing resources (e.g.:

servers, storage, databases, networking, software)
over the internet, often referred to as “the Cloud”.

 Instead of owning and maintaining physical

infrastructure, users can access and pay for these
resources on-demand.
25
Cloud Reference Model
(proposed by NIST)

26
Defining Clouds: There are many views for
what is cloud computing?
 According to NIST (National Institute of Standards and Technology):

“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,
application, and services) that can be rapidly provisioned and released with minimal
management effort or service provider interaction.”

 Buyya’s Scientific definition of Cloud Computing:

“Cloud is a market-oriented distributed computing system consisting of a collection of

inter-connected and virtualized computers that are dynamically provisioned and presented as
one or more unified computing resources based on service-level agreements (SLAs) established
through negotiation between the service provider and consumers.”

 SLA = {negotiated and agreed QoS parameters + rewards + penalties for violation of
agreement....}
27
Desired Features of a Cloud (When
Should I say service is a Cloud?)

28
Cont.
 Rapid elasticity and Scalability: These are the ability to
expand or scale-down and up the IT resources for the time
required and then release in order to meet the on-demand
requests. The resources can be scaled either horizontally or
vertically to optimize the effectiveness of the cloud datacenter
and avoid the consumers’ bottlenecks.
 On-demand computing and Self-service: This
characteristic allows the users to access the resources of
Cloud on-demand to provision, monitor and manage without
service provider intervention.
 Resource pooling: Cloud computing enables users to access
its services from a reservoir of commoditized resources (e.g.,
servers, storage, networks, applications, and services)
ubiquitously.
29
Cont.
 Broad network access. All the cloud services are made
available to its consumers through a vast network and
heterogeneous resources. The most important feature of Cloud
computing is its ubiquity nature. That means consumers can
access services at any time and from anywhere through the
internet.
 Metering (Pricing policy). All its services are offered based on
a pricing policy or metered way called “Pay-as-you-go” or “Pay-
per-use” subscription model. It is more like other utilities such as
water, electricity, telephony, gas, etc., where users are charged
based on their usage.
 Dynamic provisioning. This concept is as analogous as that of
scalability. The cloud resources or infrastructure is provisioned
when there is a spike in demand and then de-provisioned. It is
like leasing the services or resources for the time required and
then released. 30
Elasticity vs Scalability
Elasticity in Cloud Computing
 Definition:
Elasticity refers to the ability of a cloud system to automatically grow or
shrink computing resources (CPU, memory, storage, etc.) based on real-time
demand.
 Key Idea: Automatic and real-time adjustment of resources.

 Goal: Handle workload spikes efficiently and release resources when no longer
needed, optimizing cost and performance.
 Real-life Example:
 Electricity Usage in a House:
During the day, you might use more electricity (ACs, lights, appliances). At night, the
usage drops. You don’t buy a new power plant; the electricity supply adjusts
automatically to your needs—that’s elasticity.
 E-commerce Website During Sale:
During events like a Diwali or Black Friday sale, the website receives a spike in user
traffic. Cloud infrastructure automatically scales up to handle the load, and then
scales down when traffic returns to normal.
31
Cont.
Scalability in Cloud Computing
 Definition:
Scalability is the ability of a system to handle increased workloads by adding
more resources — either manually or automatically.
 Key Idea: System can grow over time (vertically or horizontally) to meet future
demand.
 Types:
 Vertical Scaling (Scale-Up): Add more power (CPU, RAM) to an existing server.
 Horizontal Scaling (Scale-Out): Add more servers to the system.

 Goal: Design a system that can grow without performance degradation.

 Real-life Example:
 Hiring More Employees in a Growing Business:
As your company grows, you hire more employees to handle the increasing
workload — similar to horizontal scalability.
 Upgrading to a Bigger Laptop:
If your current laptop is slow for your tasks, you buy one with a faster processor
and more RAM — this is vertical scalability. 32
Aspect Elasticity Scalability

Ability to automatically add or Ability to increase capacity

Definition remove resources based on real- (manually or automatically) to
time demand. handle growing workload over time.

Focus Real-time resource optimization. Long-term capacity planning.

Fluctuating workload (e.g., sudden Growing or consistently increasing

Trigger
traffic spike). workload.

Ensure system can handle more

Cost efficiency and performance
Goal load without degrading
during variable demand.
performance.

Timeframe Short-term, instant response. Long-term, planned growth.

Automation Always automatic. Can be manual or automatic.

Hiring more staff or upgrading

Electricity usage adjusting
Real-Life Example infrastructure as your company
throughout the day.
grows.

Designing a system that can support

E-commerce site auto-scaling during
Cloud Example future expansion (e.g., more users,
a flash sale.
data). 33
Cloud Computing
Architecture

34
Cont.
Bottom Layer: System Infrastructure
 Cloud Resources: Physical hardware such as servers, disks, and networking
components.
 Acts as the foundation for cloud services.
 Managed by Core Middleware to enable virtualization and abstraction.

Infrastructure as a Service (IaaS)

 Core Middleware Layer:
 Responsible for VM (Virtual Machine) provisioning, deployment, and resource
pooling.
 Abstracts physical resources using virtualization technologies.

 Cloud Hosting Platforms:

 Provides services like:
 QoS Negotiation
 SLA Management
 Metering, Accounting
 Execution & Monitoring
35
 Helps manage the lifecycle of VMs and services deployed over them.
Cont.
Category 1: Full IaaS Providers
•Provide both the management layer and the physical infrastructure
•Offer end-to-end resource provisioning and control
•Examples:
•Amazon EC2
•Microsoft Azure IaaS
•Google Compute Engine
Category 2: IaaS (M) Providers
•Offer only the management layer
•Do not own physical infrastructure
•Integrate with third-party IaaS providers
•Add value through orchestration, optimization, and unified interfaces
•Examples:
•RightScale
•Scalr
•OpenNebula (when deployed without dedicated hardware)

36
Cont.
User-level Middleware through PaaS:
 Supports application developers with platforms for Web 2.0 and Distributed
Programming (e.g., workflows, scripting)
 Cloud Programming Environment and Tools:
 Provides APIs, frameworks, and development environments.
 Developers can build and run their applications without managing underlying
infrastructure.
 Examples: Google App Engine, Microsoft Azure PaaS, Aneka

Cloud Applications through SaaS:

 Applications directly used by end users.
 Delivered over the internet, often via browser.
 Includes:
 Social Computing (e.g., Facebook)
 Enterprise ISVs
 Scientific computing tools
 Content Delivery Networks (CDNs)
37
 Examples: Google Docs, Salesforce, Dropbox
Cloud service model

 Infrastructure/Hardware as a  Service Model

Service (IaaS)
Software as a Service (SaaS)
CPU, Storage: [Link],
GoGrid….
 Platform as a Service (PaaS)
Platform as a Service (PaaS)
Google App Engine, Microsoft
Azure, Manjrasoft Aneka..
 Software as a Service (SaaS)
Infrastructure as a Service (IaaS)
[Link]

38
Manjrasoft
Infrastructure/Hardware as a
Service (IaaS)

39
Cont.
Top Layer: Web-Based Management Interface
 Interfaces: Web services, portals, REST APIs
 Purpose: Provides access to cloud resources via user interfaces or
programmable endpoints.
 Technology: Based on Web 2.0 for full-featured management consoles.
 Users: Can be humans (admin portals) or programs (automated scripts).

Bottom Layer: Physical Infrastructure

 The actual hardware and base cloud resources:
 Datacenter: Centralized resource facility consisting of thousands of nodes.
 Cluster: A group of tightly-coupled computers often used in private cloud
deployments.
 Desktop / Heterogeneous Resources: Aggregated idle computing resources
from various machines (desktop grids, lab PCs).
 Third-Party IaaS Cloud: External providers like Amazon EC2, integrated via 40
the provisioning module.
Cont.
Middle Layer: Infrastructure Management Software
 This layer manages virtual infrastructure and offers cloud orchestration
capabilities:
 Scheduling:
 Core functionality to allocate virtual machines (VMs) efficiently.
 Ensures optimal utilization of physical or external resources.

 Pricing / Billing:
 Calculates costs based on resource usage.
 Enables pay-per-use billing models for customers.

 QoS / SLA Management:

 Ensures that Service-Level Agreements are honored.
 Monitors performance and quality guarantees for VMs and services.

 Monitoring:
 Tracks VM execution and system performance.
 Helps detect failures and performance bottlenecks.

41
Cont.
•
Reservation:
•Stores future and past VM execution schedules.
•Allows for advance booking of computing capacity.
•VM Image Repository:
•Contains pre-configured VM templates (e.g., LAMP stacks).
•Users can also upload custom VM images.
•VM Pool Management:
•Maintains the status and availability of all running VM instances.
•Supports load balancing and instance health checking.
•Provisioning:
•Interfaces with third-party IaaS or physical resources.
•Deploys and manages VMs on external infrastructures when needed.

Interaction Flow
 Users or applications use web-based interfaces to interact with the cloud.

 Management software handles scheduling, VM orchestration, monitoring, and

billing.
 VMs are deployed either on local physical resources or third-party IaaS
providers.
 Dynamic provisioning supports scalability and elasticity on demand.
42
Platform as a Service (PaaS)

43
Cont.
Web-Based Interface
 Technologies: Web Services, Portals, REST API.
 Purpose: Provides users (developers/admins) access to the cloud platform.
 Use Case: Enables application deployment, configuration, and monitoring through
browser-based dashboards or APIs.

Programming API / Libraries

 Offers development tools and frameworks for building cloud-native applications.
 Can support multiple programming languages (e.g., Java, Python, .NET).
 Developers use these libraries to integrate with platform services and deploy apps.

Infrastructure Layer
 Physical Infrastructure: Data centers, clusters, and bare-metal servers managed
internally or externally.
 IaaS Providers:
 Underlying virtualized infrastructure (e.g., Amazon EC2, Microsoft Azure IaaS).
 PaaS platforms use APIs to deploy and manage workloads on IaaS layers. 44
Cont.
PaaS Core Middleware
 This is the heart of the PaaS platform, supporting essential platform
services:
 Application Management:
 Core function of the middleware.
 Automates deployment, scaling, and configuration of applications.
 Includes provisioning of databases, load balancers, and services.

 Runtime:
 Provides the execution environment for applications.
 Abstracts away infrastructure concerns like OS and VM configuration.

 Elasticity & Scaling:

 Auto-scales applications based on demand.
 Dynamically adds/removes computing resources to meet SLAs.

 Resources Management:
 Tracks, allocates, and monitors computing resources such as CPU, memory, and
I/O.
45
 Ensures fair distribution among multiple applications.
Cont.

 User Management:
 Handles user roles, permissions, and access control.

 QoS / SLA Management & Billing:

 Monitors performance against defined service levels.
 Enables billing based on resource consumption and SLA compliance.

 Pure PaaS vs Full Stack PaaS:

 Pure PaaS: Provides only the middleware (e.g., Aneka).
 Full PaaS: Includes both middleware and infrastructure (e.g., Google App
Engine).

46
Software as a service (SaaS)
SaaS is a software delivery model where applications are hosted centrally and delivered over the Internet.
•Access Model:
•Users access applications via web browsers—no need to install or maintain software locally.
•Access is typically granted via subscription or rental (recurring fee).
•Deployment:
•Applications are deployed from centralized datacenters.
•Supports delivery over Internet, Intranet, LAN, or VPN.
•Key Characteristics:
•Centrally managed applications
•One-to-many model (shared across multiple users)
•Multitenancy: Enables efficient resource utilization by serving many customers with a single instance.
•Configurable and often customizable based on user needs.
 User Benefits:
 No up-front costs for software or hardware.
 Pay-as-you-go (PAYG) pricing.
 Rapid implementation and ease of access.
 Reduced total cost of ownership (TCO).
 Automatic updates and maintenance by the provider.
47
Cont.
•Examples of SaaS Applications:
•CRM (e.g., [Link])
•ERP (e.g., NetSuite)
•Office Automation (e.g., Google Docs, Zoho Office)
•Social Networking (e.g., Facebook, LinkedIn)
•SaaS 2.0:
•Focused more on business objectives and robust infrastructure.
•Emphasizes service integration, custom component injection, and workflow design.
•Powered by Web 2.0 technologies, turning browsers into full-featured application environments.

48
Aspect SaaS 1.0 SaaS 2.0
Business outcomes and
Focus Basic application delivery
extensibility
Scalable, secure, cloud-
Infrastructure Basic hosting
native
Open APIs, flexible
Integration Limited or rigid
integrations
Custom components, logic
Customization One-size-fits-all
injection
Dynamic, visual workflow
Workflow Static
design
Web 2.0 (AJAX, JS
Tech Base Web 1.0 (HTML/CSS)
frameworks, REST)
49
Responsibilities of each stake holder

50
Cont.

51
Cloud Deployment Models

Public/Internet Private/Enterprise Hybrid/Inter

Clouds Clouds Clouds

3rd party, Mixed usage of

Cloud model run
multi-tenant Cloud private and public
within a company’s
infrastructure Clouds: Leasing public
own Data Center /
& services: cloud services
infrastructure for
when private cloud
internal and/or
* available on capacity is
partners use.
subscription basis insufficient

52
Model Hosted By Used By Key Benefit

Cost-effective,
Public Cloud 3rd Party General Public
scalable
Secure,
Private Cloud Organization Internal Use
controlled
Flexible,
Internal +
Hybrid Cloud Combination scalable +
Public
secure

53
Cont.

54
Public Cloud
 Definition: Cloud infrastructure and services are owned and operated by
third-party providers and shared among multiple users (tenants) over the
internet.
 Examples: Amazon Web Services (AWS), Microsoft Azure, Google Cloud
Platform (GCP)
 Pros:
 Low cost (pay-as-you-go)
 High scalability and elasticity
 No maintenance (provider handles infrastructure)
 Easy setup and access

 Cons:
 Less control over data and infrastructure
 Data security and compliance concerns
 Shared resources can pose performance risks

55
Private Cloud
 Definition: Cloud infrastructure is operated exclusively for a single
organization—either on-premises or hosted by a third party.
 Examples: VMware vSphere, OpenStack private deployments
 Pros:
 Greater control and customization
 Enhanced security and data privacy
 Better compliance with regulatory standards

 Cons:
 High cost of setup and maintenance
 Limited scalability compared to public cloud
 Requires in-house expertise

56
Cont.

57
Hybrid Cloud
 Definition: Combines public and private clouds, enabling data and
applications to move between them for better flexibility and optimization.
 Examples: AWS Outposts, Azure Stack, Google Anthos
 Pros:
 Flexibility to run workloads in the best environment
 Improved disaster recovery and business continuity
 Cost-effective for variable workloads

 Cons:
 Complex to manage and integrate
 Requires robust network connectivity
 Potential security loopholes during data transfer

58
Cont.

59
Community Cloud
 Definition: The infrastructure is shared by several organizations and
supports a specific community that has shared concerns (e.g., mission,
security requirements, policy, and compliance considerations). It may be
managed by the organizations or a third party and may exist on premise or
off premise.
 Examples: Government agencies, healthcare research consortiums
 Pros:
 Shared infrastructure reduces costs
 Collaboration among organizations
 Tailored security and compliance

 Cons:
 Limited availability and customization
 Potential for resource contention
 Governance complexity

60
61
Cont.

62
Pros of Cloud Computing
How Cloud computing addressed the limitations of
Cluster computing and Grid Computing?

 Resource Management: Automated and centralized

resource allocation eliminates manual configuration
complexities.

 Scalability: Provides near-infinite scalability with on-

demand resource provisioning, unlike grid and cluster's
physical infrastructure limits.
63
Cloud Computing (contd.)
 Cost Efficiency: Operates on a pay-as-you-go model,
reducing upfront investment and maintenance costs.

 Fault
Tolerance: Built-in redundancy and failover
mechanisms ensure high reliability and availability.

 Performance: Mitigates latency through edge

computing and delivers high computational power for
intensive tasks.
64
Cloud Computing (contd.)
 Ease of Use: Abstracts infrastructure complexities
with user-friendly interfaces and APIs.

 Flexibility: Supports diverse applications and provides

models like IaaS, PaaS, and SaaS to meet different
needs.

 Security: Offers advanced, centralized security

features like encryption and access management.
65
Challenges and Risks
Limitations:
 Securityand Privacy Concerns: Sensitive data
stored on cloud servers can be vulnerable to
breaches or unauthorized access. Compliance
with regulations (e.g.: GDPR, HIPAA) can be
challenging.
 Downtime and Reliability: Cloud services are
dependent on internet connectivity, and
outages can disrupt operations.
66
Cloud Computing (contd.)
 Limited Control: Users have limited control over
the infrastructure since it's managed by the
provider, which can be restrictive for certain
use cases.
 Vendor Lock-In: Migrating from one cloud
provider to another can be difficult due to
differences in platforms, APIs, and pricing
structures.
 Performance Variability: Shared infrastructure in
public clouds can lead to unpredictable
performance during peak usage. 67
Cloud Computing (contd.)
 Cost Management: While cloud computing eliminates
upfront infrastructure costs, inefficient use of
resources or unmonitored scaling can lead to
unexpectedly high bills.
 Data Latency: For real-time applications, delays in
transferring data to and from the cloud can be a
bottleneck. (physical distance between the user and
cloud data centers)
 Complexity in Integration: Integrating cloud services
with existing on-premises systems or multiple cloud
services can be challenging. 68
Major Challenges of Cloud

69
Cloud Computing Platforms

70
Research
Challenges in
Cloud
Computing

71