A small intro on distributed computing.

1. DISTRIBUTED
COMPUTING
Prepared by:
Himali A. Parikh
3rd CO

2. CONTENTS
1. Introduction
2. History
3. Overview
4. Examples of distributed systems
5. Architecture
6. Algorithms
7. Bully Algorithm : In brief
8. Fallacies of distributed computing
9. Problems related to distributed computing
2

3. INTRODUCTION
Distributed computing:
A field of computer science that studies distributed
systems.
Distributed system:
 Consists of multiple autonomous computers
that communicate through a computer
network
 Use of distributed systems to solve
computational problem
3

4. HISTORY
The first widespread distributed systems
were local-area networks such as Ethernet
that was invented in the 1970s.
E-mail is the earliest example of a large-
scale distributed application.
The study of distributed computing
became its own branch of computer
science in the late 1970s and early 1980s. 4

5. OVERVIEW
Early computing
was performed on a
single processor.
Uniprocessor
computing can be
called centralized
computing.
5

6. A distributed system is a
collection of Independent
computers, interconnected
via a network, capable of
collaborating on a task.
6

7. EXAMPLES OF DISTRIBUTED SYSTEMS
Internet
ATM machines
Intranets/Workgroups
7

8. ARCHITECTURE
Various software and hardware architectures used
for distributed computing are as follows:
Client-server
3-tier
 N-tier
Tightly-coupled
Loosely-coupled
Peer-to-peer
8

9. Client-server architecture
Advantages:
Centralization
Scalability
Flexibility
Interoperability
Disadvantages:
Dependability
Can have a single point of failure.
Server can get overloaded.
Generally more expensive and difficult
to set up initially
9

10. ALGORITHMS
1. Bully Algorithm- Leader election
2. Byzantine Fault Tolerance
3. Algorithms based on Clock synchronisation
4. Lamport Ordering
5. Algorithms based on Mutual exclusion
6. Snapshot algorithm
7. Algorithms based on Detection of process
termination
8. Vector clocks
10

11. Leader Election
In distributed computing, leader election is the process of
designating a single process as the organizer, coordinator,
initiator or sequencer of some task distributed among several
computers .
Why Leader Election is required?
The existence of a centralized controller greatly
simplifies process synchronization
However, if the central controller breaks down, the
service availability can be limited
The problem can be alleviated if a new controller
(leader) can be chosen
11

12. BULLY ALGORITHM
The bully algorithm is a method in distributed computing for
dynamically selecting a coordinator by process ID number.
The Bully Algorithm was devised by Garcia-Molina in 1982.
In this algorithm, the highest-numbered process becomes
coordinator.
Thus the biggest guy in town always wins, hence the name
“Bully Algorithm.”
12

13. When a process notices that the coordinator is not responding to
requests, it initiates an election.
Bully algorithm makes note of time of by which the process should
respond.
Election is held as follows:
–P sends an ELECTION message to all processes with higher
numbers.
–If no one responds, P wins the election and becomes coordinator.
–If one of the higher-ups answers, it takes over. P’s job over!
13

14. 1.Process 4 holds an election.
2.Process 5 and 6 respond, sending
OK message that , we are UP.
3.Now 5 and 6 each hold an
election
14

15. 4.Process6 tells 5 to stop
5.Process 6 wins and tells everyone
15

16. Fallacies of distributed computing
1. The network is reliable.
2. Latency is zero.
3. Bandwidth is infinite.
4. The network is secure.
5. Topology doesn't change.
6. There is one administrator.
7. Transport cost is zero.
8. The network is homogeneous.
16

17. Benefits
Resource sharing
Scalability
Fault tolerance
Availability
Performance
Challenges
Heterogeneity
Latency
Need for “ OPENNESS”
Scalability
Transparency
17

18. CONCLUSION
18

19. THANK YOU
19