Nothing Special   »   [go: up one dir, main page]

UNIT 6 Hardware & Software Concepts PDF

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Distributed Systems:

Hardware and Software Concepts


Gian Pietro Picco
Dipartimento di Elettronica e Informazione
Politecnico di Milano, Italy
picco@elet.polimi.it
http://www.elet.polimi.it/~picco

P olitecnico
di M ilano

Distributed Hardware

Multiprocessors vs. multicomputers


Homogeneous vs. heterogeneous (multicomputers)
Parallel computing vs. distributed computing
Distributed Systems - 2004-2006 G.P. Picco

P olitecnico
di M ilano

Classes of Distributed Systems


System

Description

Main Goal

Distributed OS

Tightly-coupled operating system for multiprocessors and homogeneous multicomputers

Hide and manage


hardware resources

Network OS

Loosely-coupled operating system for


heterogeneous multicomputers (LAN and WAN)

Offer local services


to remote clients

Middleware

Additional layer atop of NOS implementing generalpurpose services

Provide distribution
transparency

All provide support for the development of distributed


applications
The first two also provide services directly to users
Distributed Systems - 2004-2006 G.P. Picco

P olitecnico
di M ilano

Distributed Operating System

Multiprocessor vs. multicomputer


No physical shared memory: only message passing
Different semantics, different reliability assumptions
Distributed Systems - 2004-2006 G.P. Picco

P olitecnico
di M ilano

Distributed Shared Memory


Powerful abstraction
for programmers
Nothing changes w.r.t.
centralized models
No need for message
passing

Optimized using
replication, to
increase availability
Efficiency problems
False sharing

Distributed Systems - 2004-2006 G.P. Picco

P olitecnico
di M ilano

Network Operating System

BSD sockets, NFS are examples of services


At the core of the client-server paradigm
Distributed Systems - 2004-2006 G.P. Picco

P olitecnico
di M ilano

Network Operating System

Users may have both explicit access to remote


resources (e.g., through scp, ssh) and implicit access
(e.g., through mounted file systems)
Inconsistent views across the system

Heterogeneity and independency of host:


Cons: difficult to manage and to use
Pro: more scalable

Programming abstraction are very low-level


Distributed Systems - 2004-2006 G.P. Picco

P olitecnico
di M ilano

Middleware

Hides the complexity of distribution behind a well-designed


programming model
Examples: CORBA, RMI, publish-subscribe, TP monitors,
Discussed in detail during the course

Distributed Systems - 2004-2006 G.P. Picco

P olitecnico
di M ilano

Comparison
Distributed OS
Item

Network OS

Middleware

Multiprocessor

Multicomputer

Degree of transparency

Very High

High

Low

High

Same OS on all nodes

Yes

Yes

No

No

Number of copies of OS

Basis for communication

Shared memory

Messages

Files

Model specific

Resource management

Global, central

Global,
distributed

Per node

Per node

Scalability

None

Moderate

High

Varies

Openness

Closed

Closed

Open

Open

Distributed Systems - 2004-2006 G.P. Picco

You might also like