1 of 50

DISTRIBUTED COMPUTING

Sunita Mahajan, Principal, Institute of Computer Science, MET League of Colleges, Mumbai

Seema Shah, Principal, Vidyalankar Institute of Technology, Mumbai University

2 of 50

Chapter-1��Basic Distributed System Concepts

Operating system

(hardware, Software, application software)

Network (Resource, Client-server)

Communication

Database Concept

Independent , networked, communicating

3 of 50

What is a distributed system?

Tanenbaum’s definition of a distributed system:

“A distributed system is a collection of independent computers that appear to the users of the system as a single coherent system.”

DS is a software system in which components located on networked computers and they communicate and coordinate by passing msg

Component interact with each other: for common goal

4 of 50

5 of 50

DS-Goals

Making resource availability
Distribution Transparency
Scalability
Openness- interact services from other open system

E.g. message (Send, Receive)

6 of 50

Principles of DS

Coordinate activities, Share Resource of system

Processes
Communication
Naming – scale tech DNS (.com .au .in)
Synchronization
Replication and consistency
Fault tolerance and security

7 of 50

Architecture for DS

Single high speed processor (centralize)
Decentralize (not in syllabus)
Multiple processors inter connected (Distributed)

(i) tightly coupled (shared Memory)

(ii) Loosely coupled (Distributed memory arch)

Compare above three?

(point of failure, Fault tolerence/stability, scalability)

8 of 50

9 of 50

Architectures for Distributed systems

10 of 50

An example of a Distributed System

Nationalized Bank with multiple Branch Offices

11 of 50

Requirements of Distributed systems

Security and reliability.
Consistency of replicated data.
Concurrent transactions (operations which involve accounts in different banks; simultaneous access from several users, etc)
Fault tolerance

12 of 50

Architectures for Distributed systems

Tightly coupled systems (Shared memory)

shared memory (address space)
System- wide primary memory -> nr. of PE’s
Write carefully
Should have exclusive access, Parallel processing

Loosely coupled systems (Distributed memory )

Do not share memory, local memory
Communication - passing msg across network
scalability

13 of 50

Architectures for Distributed systems

14 of 50

NOS vs DOS

NOS: Network OS
Software associate

(protocol)

Multiplicity of nodes

Control-manual
Performance- bad
Remote resource

login- failure

DOS: Distribute OS
Centralize OS ,

run on multiple PE

User unaware of multiplicity (machines)
Automatically
Good performance
Actual local resource

15 of 50

Distributed Computing Models

Mini computer models

(centralize time sharing sys., multiple users, comm n/w)
E.g. ARPANET

Workstation model
Workstation–server model
Processor-pool model
Hybrid model

16 of 50

Workstation model

Consists of network of Personal computers,
Each one with its own hard disk and local file system
Interconnected over the network(high speed LAN)
single user sys,Idle- when not in use (user absent), waste of cpu

WS---job --> WS (efficiency, processing capability)

Issue: how to find idle WS?, how transfer, (idle -> busy)

17 of 50

workstation-server model

Consists of multiple workstations coupled with powerful servers with extra hardware to store the file systems and other software like databases
Not own disk, file on server, (less expensive)
High speed LAN, different servers
Server control (Fun, reliability, fault tolerant)

18 of 50

processor-pool model

consists of multiple processors: a pool of processors and a group of workstations
Run Server (pool schedule)

19 of 50

Advantages of Distributed systems

Inherently distributed applications
Information sharing among geographically distributed users
Resource Sharing
Better price performance ratio
Shorter response time & higher throughput
Higher reliability and availability against component failures
Extensibility and Incremental Growth
Better Flexibility
e,.g. banking, ticket reservation,

multi branch offices

20 of 50

Disadvantages of Distributed systems

Relevant software does not exist currently
Security poses a problem due to easy access to all data
Networking saturation may cause a hurdle in data transfer.

21 of 50

Software concepts

Network Operating System (NOS)
Distributed Operating System (DOS)
Multiprocessor Time Sharing System

22 of 50

Network Operating System (NOS)

Build using a distributed system from a network of workstations connected by high speed network.
Each workstation is an independent computer with its own operating system, memory and other resources like hard disks, file system and databases

23 of 50

Distributed Operating System (DOS)

Enables a distributed system to behave like a virtual uniprocessor even though the system operates on a collection of machines.
Characteristics

enabling Inter process communication,
Uniform process management mechanism,
Uniform and visible file system,
Identical kernel implementation,
Local control of machines
handling scheduling issues.

24 of 50

Multiprocessor Time Sharing System

Combination of tightly coupled software and tightly coupled hardware with multiple CPUs projecting a uniprocessor image.
Tasks are queued in shared memory and are scheduled to be executed in time shared mode on available processors.

25 of 50

Comparison of different Operating systems

Software Concepts

26 of 50

Issues in Designing Distributed systems

Transparency
Flexibility
Reliability
Performance
Scalability
Security

27 of 50

Transparency

Transparencies required for Distributed Systems

28 of 50

Replication Transparency

Locating Replicated File stored on any server

29 of 50

Flexibility

Monolithic kernel approach
Microkernel approach

Kernel- central control (Basic system facilities)
Kernel-execute supervisory mode

Privilege access to physical resources
Function of kernel

Memory management
Process management
Resource management

30 of 50

Monolithic kernel approach

uses the minimalist , modular approach with accessibility to other services as needed.

31 of 50

Microkernel approach

uses the kernel does it all approach with all functionalities provided by the kernel irrespective whether all machines use it or not

32 of 50

Monolithic versus Microkernel Approach

33 of 50

�Reliability�

Availability in case of Hardware failure

- e.g. single system- if (crash)?

- multiple proc available, become reliable

Data recovery in case of Data failure
Maintain consistency in case of replicated data

34 of 50

Performance

Metrics are:

Response time,
Throughput,
System utilization
Amount of network capacity used

35 of 50

Scalability

Techniques to handle scalability issues

hide communication latencies,
hide distribution
hide replication

36 of 50

Hide communication latencies,

37 of 50

Hide distribution

38 of 50

Security

confidentiality means protection against unauthorized access;
integrity implies protection of data against corruption
availability means protection against failure always accessible.

Fault tolerence

Fail-> other one take load

39 of 50

Client Server model

Client Server Interaction

40 of 50

Client Server addressing techniques

Machine addressing,
process addressing
Name server addressing

41 of 50

Client Server addressing techniques

42 of 50

Client Server implementation

Messages for client server interaction

Request, Reply, Acknowledge, Are you Alive, I am Alive.

43 of 50

differentiation between the client and the server

User interface level
Processing level
data level

44 of 50

Client Server Architecture

45 of 50

46 of 50

Case Study: World Wide Web 1.0

47 of 50

Internet scenario with web servers and web browsers

48 of 50

Case study: World Wide Web 2.0

49 of 50

Case Study: Google Servers

50 of 50

Summary

Multiple-interconnected computers can have either shared memory or distributed memory architectures
Distributed systems offer integration of distributed applications, resource sharing, more reliability, better flexibility
Network operating system, distributed operating system and multiprocessor timesharing system are different types of distributed systems