CSE 461: Cloud Computing

Lecture 2
Introduction to Cloud Computing
Prof. Mamun, CSE, HSTU
 Lecture Outline
Discussion On

Software Service
Models
Types of Clouds
and A Private
The Cloud Stack Cloud
Service Models
of Cloud
Cloud Computing
Infrastructure

2
 Lecture Outline
Discussion On

Software Service
Models
Types of Clouds
and A Private
The Cloud Stack Cloud
Service Models
of Cloud
Cloud Computing
Infrastructure

3
 What is a Server?
 Servers are computers that provide “services” to “clients”

 They are typically designed for reliability and to service a large

number of requests

 Organizations typically require many physical servers to provide

various services (Web, Email, Database, etc.)

 Server hardware is becoming more powerful and compact

4
 Compact Servers
 Organizations would like to conserve the amount of floor space
dedicated to their computer infrastructure

 For large-scale installations, compact servers are used. This helps with:
 Floor Space
 Manageability
 Scalability
 Power and Cooling

5
 Racks
 Equipment (e.g., servers) are typically placed in racks

 Equipment are designed in a modular fashion to fit into rack units

(1U, 2U etc.)

 A single rack can hold up to 42 1U servers

1U Server

7U Blade center

6
 Blades and Blade Enclosures
 A blade server is a stripped down computer with a
modular design

 A blade enclosure holds multiple blade servers and

provides power, interfaces and cooling for the individual
blade servers

7
 Blade Performance
 Consider bandwidth and latency between these layers
Quad Core Processor

Disk
core core

core core
L1

L2

Quad Core Processor

core core L3

core core
L1
Memory
L2

8
 …Performance across blades
 Consider bandwidth and latency across blades

Network is usually the bottleneck

9
 What is a Data Center?
 A data center is a facility used to house computer systems and
associated components, such as networking and
storage systems, cooling, uninterruptable power supply, air filters…

 A data center typically houses a large number of heterogeneous

networked computer systems

 A data center can occupy one room of a

building, one or more floors, or an
entire building

10
 Data Center Components
 Air conditioning
 Keep all components in the manufacturer’s recommended
temperature range
 Redundant Power
 UPS/Generators
 Multiple power feeds
 Fire protection
 Physical security
 CCTV/Access Control
 Monitoring Systems
 Connectivity
 Multiple ISPs/Leased Lines

11
The Network of a Modern Data Center
Internet
CR CR
Layer 3
AR AR AR AR

LB S S LB …

Layer 2
S S S S

A A … A A A … A

 CR = L3 Core Router, AR = L3 Access Router, S = L2 Switch, LB = Load Balancer,

A = Rack of 20 servers (Cisco with ~ 4,000 servers) 12
Communication In Data Centers
 Communication in data centers are most often based on networks running
the IP protocol suite

 Data centers contain a set of routers and switches that transport traffic between
the servers and to the outside world

 Traffic in today’s data centers:

 80% of the packets stay inside the data center
 Trend is towards even more internal communication

 Typically, data centers run two kinds of applications:

 Outward facing (serving web pages to users)
 Internal computation (data mining and index computations–think of
MapReduce and HPC)

13
 Communication Latency
 Propagation delay in the data center is essentially 0
 Light goes a foot in a nanosecond

 End to end latency comes from

 Switching latency
 10G to 10G:~ 2.5 usec (store&fwd); 2 usec (cut-thru)
 Queuing latency
 Depends on the size of queues and network load

 Typical times across a quiet data center: 10-20usec

14
 Elasticity and Performance
 Bare data centers make it hard for applications to grow/shrink
 VLANs can be used to isolate applications from each other
 IP addresses are topologically determined by
Access Routers

 Reconfiguration of IPs and VLAN trunks is painful, error-

prone, slow, and often manual

 In addition, no performance isolation is provided:

 VLANs typically provide reachability isolation only
 One service sending/receiving too much traffic hurts all
services sharing its subtree

15
 Power in Data Centers
 Pretty good data centers have efficiency of 1.7
 0.7 Watts lost for each 1W delivered to the servers

 How can we reduce power costs?

 Create servers that use less power?
 Conventional server uses 200 to 500W
 Reductions have ripple effects across entire data center
 Mostly a problem for scientists to tackle!!

 Eliminate power redundancy?

 Allow entire data centers to fail

 Reduce power usage of network gear?

 Total power consumed by switches amortizes to 10-20W per server

16
 Utilization In Data Centers
 Utilization of 10% to 30% is considered “good” in data centers

 Causes:
 Uneven application fit:
 Each server has CPU, memory, and disk: most applications
exhaust one resource, stranding the others
 Long provisioning timescales
 Uncertainty in demand:
 Demand for a new service can spike quickly
 Risk management:
 Not having spare servers to meet application demands leads
to failure

17
 What About?
 Maximize useful work per dollar spent – 59% of dollars are spent on
servers with very low utilization (10%)

 Turn the servers into a single large resource pool and let services
“breathe” : dynamically expand and contract their footprint
as needed

 Two main requirements: Enabled by Virtualization

 Means for rapidly and dynamically satisfying application
fluctuating resource needs
Enabled by Programming Models and Distributed File Systems
 Means for servers to quickly and reliably access shared and
persistent data
 Data too large to copy during provisioning process

18
 A Cloud is …
 A data center hardware and software that the vendors use to offer
the computing resources and services

19
 Cloud Computing
Cloud Computing is the
delivery of computing as a
service rather than a
product,

whereby shared
resources, software, and
information are provided to
computers and other
devices,

as a metered service over

a network.
 Lecture Outline
Discussion On

Software Service
Models
Types of Clouds
and APrivate
The Cloud Stack Cloud
Service Models
of Cloud
Cloud Computing
Infrastructure

21
 IT as a Service
 How do you offer IT as a service?
 Different users have different needs
 Consider the needs of:
 Average End User
 Application Developer
 Enterprise System Architect

Let us look at some of the typical service models

22
Cloud Service Models

• Software-as-a-Service

SaaS
• Applications running on browsers

• Platform-as-a-Service

PaaS • A software platform that is made

available to developers to build
cloud applications

• Infrastructure-as-a-Service

IaaS • Basic computing resources such

as CPU/Memory/Disk, made
available to users in the form of
Virtual Machine Instances

23
 SaaS
 You are most familiar with this! SaaS

PaaS

 Software is delivered as a service IaaS

over the Internet, eliminating the

need to install and run the
application on the customer's
own computer

 This simplifies maintenance and

support

 Examples: Google Workspace

(previously known as G Suite),
Microsoft 365, Dropbox, Slack,
Zoom, Adobe Creative Cloud,
and Salesforce, among others 24
 SaaS Maturity Levels
 Distinguishing attributes: configurability, multi-tenant
efficiency, scalability

1 2 3 4

Tenant 1 Tenant 2 Tenant 1 Tenant 2 Tenant 1 Tenant 2 Tenant 1 Tenant 2

Tenant Load Balancer

instance 1 instance 2 instance instance instance

instance instance
Configurable + Multi-tenant-efficient + Scalable
• Each has its own • Same application but •(+):Efficient use of server
customized version of distinct instance/customer resources without apparent
the application and run differences to end users
its own instance • (-): scalability limits

25
 PaaS
 The Cloud provider exposes a set of SaaS
tools (a platform) which allows users to PaaS
create SaaS applications
IaaS

 The SaaS application runs on the

provider’s infrastructure

 The cloud provider manages the

underlying hardware and requirements

26
 PaaS Example I
 Google App Engine

Build web applications on Google’s Infrastructure

27
 PaaS Example II
 The Facebook Developer Platform

Set of APIs that allow you to create Facebook Applications

28
 IaaS (1/3)
 The cloud provider leases to users SaaS

Virtual Machine Instances (i.e., PaaS

computer infrastructure) using the IaaS

virtualization technology

OS OS
 The user has access to a standard VM VM

Operating System environment and Hypervisor

can install and configure all the

layers above it Server

Server

Server

Server

Cloud Provider

29
 IaaS (2/3)
SaaS
 The virtualization technology is a major PaaS

enabler of IaaS IaaS

HARDWARE 31
 IaaS (3/3)
Capacity
Service Request Operations Dynamic
Monitoring Planning
Catalog UI UI Scheduling
SLA

Request Driven Provisioning & Service Management

Web 2.0 Data

Software Virtual High Volume
Collaborative Intensive
Development Classroom Transactions
Innovation Processing

Workloads

Virtual Virtual Virtual Virtual Virtual

Servers Storage Networks Applications & Clients
Virtualization Middleware

Servers Power Systems Racks, BladeCenter Storage Networking

Physical Layer

31
 IaaS Example
 Amazon Web Service Elastic Compute Cloud (EC2)

32
 Other Service Models
 Hardware-as-a-Service
 Communication-as-a-Service
 XaaS
 “X” as a Service

33
 Datacenter-as-a-Service
 Increasing Number of Servers

 Manpower, Electricity, Cooling, Security?

 Management Nightmare

 Why not give it to someone else?

34
 Lecture Outline
Discussion On

Software Service
Models
Types of Clouds
and A Private
The Cloud Stack Cloud
Service Models
of Cloud
Cloud Computing
Infrastructure

35
The Cloud Stack
Applications
Data
Runtime
Middleware
Operating System
Virtualization
Servers
Storage
Networking

36
 Applications
 Cloud applications can range
from Web applications to
Applications scientific computational jobs
Data
Runtime
Middleware
Operating System
Virtualization
Servers
Storage
Networking

37
 Data
 Data Management
 New generation of cloud-
Applications
Data
specific databases and
Runtime
Middleware
management systems
 E.g., Hbase, Cassandra,
Operating System
Virtualization
Servers
Storage
Networking
Hive, Pig etc.

38
 Runtime Environment
 Runtime platforms to support
cloud programming models
Applications
Data
 E.g., MPI, MapReduce,
Runtime
Middleware
Pregel etc.
Operating System
Virtualization
Servers
Storage
Networking

39
 Middleware for Clouds
 Management platforms that
enable:
Applications  Resource Management
Data
Runtime
Middleware
 Monitoring
Operating System
Virtualization
 Provisioning
Identity Management and
Servers
Storage 
Security
Networking

40
 Operating Systems
 Standard Operating Systems
used in Personal Computing
 Packaged with libraries and
software for quick deployment
Applications
Data

and provisioning
Runtime
Middleware

 E.g., Amazon Machine Images

Operating System
Virtualization

(AMI) contain OS as well as

Servers
Storage
Networking
required software packages as
a “snapshot” for instant
deployment

41
 Virtualization
 Key Component
 Resource Virtualization
Applications
Data  Amazon EC2 is based on the
Xen virtualization platform
Runtime
Middleware
Operating System
Virtualization
Servers
Storage
Networking

42
 Cloud Service Layers in the
Service Levels SaaS

PaaS

IaaS

Packaged IaaS PaaS SaaS

Software

User Managed
Applications Applications Applications Applications
User Managed

Data Data Data Data

Runtime Runtime Runtime Runtime

Vendor Managed
User Managed

Middleware Middleware Middleware Middleware

Vendor Managed
Operating System Operating System Operating System Operating System
Vendor Managed

Virtualization Virtualization Virtualization Virtualization

Servers Servers Servers Servers
Storage Storage Storage Storage
Networking Networking Networking Networking

43
 Lecture Outline
Discussion On

Software Service
Models
Types of Clouds
and A Private
The Cloud Stack Cloud
Service Models
of Cloud
Cloud Computing
Infrastructure

44
 Types of Clouds (1/4)
 Public
 Private
 Hybrid

45
 Types of Clouds (2/4)
 Public (external) cloud
 Open market for on demand computing and IT resources
 Concerns: Limited SLA, reliability, availability, security,
trust and confidence
 Examples: IBM, Google, Amazon, …

46
 Types of Clouds (3/4)
 Private (Internal) cloud
 For enterprises/corporations with large
scale IT

47
 Types of Clouds (4/4)
 Hybrid cloud
 Extend the private cloud(s) by connecting it to other external
cloud vendors to make use of their available cloud services

 Cloud Burst
 Use the local cloud, and when you need more resources,
burst into the public cloud

48
 An Example:
A Private Cloud
 Qloud Stack
Bayesian Classification, K-Means, etc.
Applications
Data HDFS

Runtime Apache Hadoop 0.20.1

Middleware Ganglia cluster monitoring system and the VMware vSphere Client

64-Bit Fedora 13
Operating System
Vmware vSphere 4.1/ESXi 4.1
Virtualization
14 IBM Quad Core (E5420) Blades
Servers
Storage/Blade = 2 x 300 GB SAS & RAM/Blade = 8 GB RAM
Storage
Networking

50
 Lecture Outline
Discussion On

Software Service
Models
Types of Clouds
and A Private
The Cloud Stack Cloud
Service Models
of Cloud
Cloud Computing
Infrastructure

51
Economics of Cloud Computing
 Evolution of Software Service Models
 What is the Value Proposition for Cloud
Computing?
 How did Cloud Computing emerge from
business / industry rather than from
Academia?

52
Cost of Information Technology
 When you are using IT there are three
primary costs associated with it:
 Software Cost (Media + License cost/user)
 Support Cost (Vendor Support, Updates and
Patches etc.)
 Management Cost (IT Infrastructure costs,
Manpower, etc.)

53
 Traditional Model
 Classical Model
 Software provider develops software and
charges a license fee per user for the client
 The provider may charge a support fee /user
 The management of the software is the
clients responsibility
 Up to 4x the cost of the actual software per year!
 Infrastructure, Manpower, software maintenance
 Traditional Software – Oracle etc.

54
 Software Service Models
Traditional

$4000 /user
Software Cost
(one-time)

Support Cost $800 /user

/year

Up to 4x the
cost of
Management Cost
Software!

Deployment
Client Side
Location

55
 Open Source Model
 “Free” Model
 Software provider packages Open Source
Software and provides it at little or no cost to
the client
 The provider makes money on support –
charges a higher fee than traditional model
 The cost of Managing the software remains
the same as Traditional Model
 Up to 4x the cost of the actual software per year!
 Infrastructure, Manpower, software maintenance

56
 Software Service Models
Open
Traditional
Source

$4000 /user
Software Cost $0 /user
(one-time)

Support Cost $800 /user $1600 /user

/year /year

Management Cost Up to 4x the cost of Software!

Deployment
Client Side
Location

57
 Outsourcing Model
 Primary cost of Software Management is
in Manpower
 Why not delegate the management of
software to a country with cheaper labor
costs
 India, China etc.
 Outsource the management of software
for a flat fee – keep IT management costs
under control

58
 Software Service Models
Open
Traditional Outsourcing
Source

$4000 /user $4000 /user

Software Cost $0 /user
(one-time) (one-time)

Support Cost $800 /user $1600 /user $800 /user

/year /year /year

< 1300 /user

Management Cost Up to 4x the cost of Software!
/month

Deployment Client or
Client Side Provider Side
Location

59
 Hybrid and Hybrid+ Model
 Business Software Requirements do not change often.
 ERP/Financials/CRM etc.
 Why reinvent the wheel?
 Standardize, Specialize and Repeat
 Create a flexible version of the Software that can be
quickly configured and deployed.
 Automate support through remote access.
 Sell easy to deploy software to many clients.
 Decrease the Margin
 Increase the Customers
 Hybrid+ is more advanced – charge a flat monthly fee
for the software, support and management

60
 Software Service Models
Open
Traditional Outsourcing Hybrid Hybrid+
Source

$4000 /user $4000 /user $4000 /user

Software Cost $0 /user
(one-time) (one-time) (one-time)

Support Cost $800 /user $1600 /user $800 /user $800 /user $300 / user
/year /year /year /year month

Bid < 1300 $150 /user

Management Cost Up to 4x the cost of Software!
/user /month /month

Deployment Client or Provider Side

Client Side
Location

61
 Software as a Service Cloud
Computing
 Develop Web Application
 Offer to customers over Internet
 No deployment costs
 Amortize Management and Support costs
over many clients

62
 Software Service Models
Open
Traditional Outsourcing Hybrid Hybrid+ SaaS
Source

$4000 /user $4000 /user $4000 /user

Software Cost $0 /user
(one-time) (one-time) (one-time)

Support Cost $800 /user $1600 /user $800 /user $800 /user $300 / user < $100 /user
/year /year /year /year month /month

Bid < 1300 $150 /user

Management Cost Up to 4x the cost of Software!
/user /month /month

Deployment Client or Provider Side Provider Side

Client Side
Location

63