4005-730 -- Distributed Systems

Alan Kaminsky

•

Department of Computer Science

•

Rochester Institute of Technology

•

4486 + 2220 = 6706

Home Page

Distributed Systems

4005-730-01

Spring Quarter 2013

Course Page

4005-730 Distributed Systems
Lecture Notes -- Module 1. Fundamentals

Prof. Alan Kaminsky
Rochester Institute of Technology -- Department of Computer Science

The Evolution of (So-Called) Distributed Systems

Mainframe batch systems (1950s)
- Input/output = punched cards and printouts
- Computation/data = mainframe
- Interconnect = direct
Mainframe time-sharing systems; single-user, multi-user (1970s)
- Input/output = dumb terminal
- Computation/data = mainframe
- Interconnect = serial line or modem
PCs (1980s)
- Input/output = graphics terminal + mouse
- Computation/data = microprocessor, disk
- Interconnect = direct
Two-tier and multi-tier systems; single-user, multi-user (1980s-1990s)
- Input/output = PC with specialized software
- Computation/data = mainframe
- Interconnect = local network
Client-server systems; single-user, multi-user (1990s)
- Input/output = PC with generalized software (web browser)
- Computation/data = server
- Interconnect = Internet
Modernized client-server systems; single-user, multi-user (2000s)
- Input/output = PC, tablet, smartphone
- Computation/data = server
- Interconnect = Internet
What many would call "distributed" systems look just like 1950s-era mainframe batch systems (with more sophisticated components), and so are not really "distributed" at all!

http://www.xkcd.com/934/

Distributed System Architectures

A three-dimensional space

Dimension 1: End-to-end communication patterns

                         One-to-many
One-to-one               Many-to-one              Many-to-many
0-----1-----2-----3-----4-----5-----6-----7-----8-----9-----10

Dimension 2: Servers

Central servers                             No central servers
0-----1-----2-----3-----4-----5-----6-----7-----8-----9-----10

Dimension 3: Address configuration

Preconfigured                                           Ad hoc
0-----1-----2-----3-----4-----5-----6-----7-----8-----9-----10

Where do various distributed applications fall along each dimension?
- World Wide Web
- Email
- Webcasting (live audio, live video)
- File sharing
- Chat room
- Multiplayer online games
Client-server systems
Peer-to-peer (P2P) systems
Ad hoc collaborative systems

Distributed System Technologies and Paradigms

Internet protocols
- Sockets -- TCP/IP
- Datagrams -- UDP/IP
Distributed objects
- Common Object Request Broker Architecture (CORBA) -- IIOP
- Java Remote Method Invocation (Java RMI) -- JRMP, IIOP
- Microsoft .NET -- SOAP
Service federations
- Jini Network Technology, Salutation (now defunct)
- Web services
Paradigms
- Message oriented systems
- Tuple spaces
- Agent systems
- Map-reduce systems
- Cloud/grid computing
- Supercomputing

Deutsch's Eight Fallacies of Distributed Computing

The Eight Fallacies of Distributed Computing

The network is reliable
Latency is zero
Bandwidth is infinite
The network is secure
Topology doesn't change
There is one administrator
Transport cost is zero
The network is homogeneous

I would add one more:

System clocks are identical

Some of these go away in a client-server setting

Distributed System Design Issues

Communication
- Getting different processes to talk to each other
- Messages
- Remote method invocation
- Tuples
Fault tolerance
- Single-CPU systems experience only total failures
- Distributed systems experience partial failures
- Failure modes: Process crashes; host crashes; network failures
- Resource recovery -- leasing
- No single point of failure -- replication
- All or nothing -- transactions, distributed transactions
Coordination
- Getting different processes to coordinate operations with each other
- Concurrency
- Mutual exclusion
- Deadlocks and anti-deadlock techniques
- Esoteric algorithms -- elections, consensus
Security
- Confidentiality
- Authentication
- Integrity
- Replays
Some of these go away in a client-server setting