distributed systems1

19
Distributed Systems By- Sumita Das

Upload: sumita-das

Post on 23-Jan-2017

42 views

Category:

Engineering


0 download

TRANSCRIPT

Page 1: Distributed systems1

Distributed SystemsBy- Sumita Das

Page 2: Distributed systems1

2

Topics Covered

Introduction to Distributed systems Examples Advantages over Personal Systems Advantages over Centralized Systems Disadvantages Design Issues Communication in Distributed Systems

Created by Sumita Das

Page 3: Distributed systems1

Created by Sumita Das 3

Distributed System

Page 4: Distributed systems1

Created by Sumita Das 4

Definition

Collection of independent computers that appears to users of the system as a single computer.

Each computer consists:•Memory •CPU

Allow many user to work together.

Page 5: Distributed systems1

Created by Sumita Das 5

Examples of distributed systems

Web search –Index the entire contents of the Web.

Massively multiplayer online games –Very large number of users sharing a virtual world.

Financial trading –Real time access and processing of a wide range of information sources

Social Networking (Facebook)

Page 6: Distributed systems1

Created by Sumita Das 6

Advantages over Personal Systems

1. Data sharing : Many users access same database.

2. Device sharing: Allow many users to share peripherals.

3. Communication: Make human-to-human communication easier.

4. Flexibility: Spread workload over available machines.

Page 7: Distributed systems1

Created by Sumita Das 7

Advantages over Centralized systems

1. Speed: Has more total computing power than mainframe.

2. Reliability: If one machine crashes, the system as a whole can still survive.

3. Incremental growth: Computing power can be added

Page 8: Distributed systems1

Created by Sumita Das 8

Advantages over Centralized systems

4. Economics: Offer better price and performance than mainframes.

5. Inherent distribution: Some machines involve spatially separated machines

Page 9: Distributed systems1

Created by Sumita Das 9

Disadvantages of Distributed System

1. Software: Little software exists at present

2. Networking: Network can saturate or cause other problems

3. Security: Easy access also applies to secret data

4. Privacy: Data may be accessed securely but without the owner’s consent

Page 10: Distributed systems1

Created by Sumita Das 10

Design Issues

Making Resources Accessible Transparency Flexibility Reliability Performance Scalability

Page 11: Distributed systems1

Communication in Distributed SystemsUniprocessor Systems

Distributed Systems

Inter Process Communication

Shared Memory

Message Passing

Remote Procedure Call

Page 12: Distributed systems1

Message Passing Model

Communication Primitives :•SEND

•Message•Destination

•RECEIVE•Source of message•Buffer for storing message

Page 13: Distributed systems1

Message Passing Model

Blocking Primitives

Non-Blocking Primitives

•The user buffer can be reused as soon as the control is returned to the user program.

•Returns control to the caller immediately before message is sent to the user.

Page 14: Distributed systems1

Created by Sumita Das 14

Disadvantage: Sender can’t modify the message buffer until the message has been sent.

Non-Blocking Primitives

Non-Blocking Primitives

Buffered Option Unbuffered Option

Page 15: Distributed systems1

Created by Sumita Das 15

Buffered Option

Messages are copied three times

1. User buffer to the kernel buffer2. Kernel buffer on the sending computer to the

kernel buffer on the receiving computer3. Buffer on the receiving computer to a user

buffer.

Page 16: Distributed systems1

Created by Sumita Das 16

Data is copied from one user buffer to another user buffer directly.

SEND : It should avoid reusing the user buffer until the message has been transmitted.

RECEIVE : If a message is present, the consumer process reads it, otherwise it performs some other computation.

Unbuffered Option

Page 17: Distributed systems1

Created by Sumita Das 17

Message Passing Model

Reliable Primitives

Unreliable Primitives

Unreliable SEND : It does not return control to the user program until the message has been sent.

Reliable SEND : It does not return control to the user program until an acknowledgment has been received.

Page 18: Distributed systems1

Created by Sumita Das 18

RECEIVE : does not return control until a message is copied to the user buffer.

Reliable RECEIVE automatically sends an acknowledgment.

Unreliable RECEIVE does not send an acknowledgment.

Message Passing Model

Page 19: Distributed systems1

Created by Sumita Das 19

References

• Andrew S.Tanenbaum, Maarten Van Steen, “DISTRIBUTED SYSTEMS: Principles and Paradigms, Second Edition