11-10-2012, 04:11 PM
Distributed Computing System Fundamentals
Distributed Computing.pptx (Size: 247.73 KB / Downloads: 29)
Introduction to Distributed Computing Systems
Advancement in microelectronic technology availability of fast, inexpensive processors
Advancements in communication technology availability of cost-effective and highly efficient computer networks.
Net Result the price-performance ratio has now changed to favor the use of interconnected , multiple processors in place of single, high-speed processor
Tightly Coupled Multiprocessor System
Also referred as Parallel Processing Systems
Single system wide primary memory
Memory is shared by all Processors
Any communication between processors usually takes place through shared memory
Disadvantages:
All processors must reside at same geographical location
Number of processors are usually small and limited by bandwidth of shared memory
Loosely Coupled Multiprocessor System
Also referred as Distributed Computing Systems OR Distributed Systems
Processors do not share memory
Every processor has its own local memory and peripherals
All communication between processors is done by passing messages across communication network that interconnects the processors
For a particular processor, its own resources are local and, other processor’s resources are remote
Together, a processor and its resources are usually referred as node or site or machine of Distributed Computing System
Advantage:
Processors of Distributed Computing Systems can be located far from each other to cover a wider geographical area
System is freely expandable and can have almost unlimited number of processors
Workstation Model
Several workstations interconnected by a communication n/w
Every workstation is a single user system
Usually connected through LAN
Idle workstation can be used by a user of a workstation as an additional computing resource
A user logs onto “home “ workstation
If home workstation finds that users terminal task needs more processing, it transforms one or more process to idle workstation in the network.
Model not easy to implement
Issues:
How does a system find an idle workstation?
How a process is transferred from one workstation to another for execution?
What happens to a remote process if user logs onto a workstation then was idle until now & was being used to execute a process of another workstation?
Hybrid Model
Workstation server model useful where large number of users perform simple interactive jobs like editing, sending and receiving emails, executing small programs.
Processor pool model useful where group of users who often perform jobs needing massive computation.
Hybrid model of workstation server and processor pool models may be used to combine the advantages.
Distributed Computing Systems
According to Tanenbaum and Van Rensee [1985], two primary tasks of OS are :
To provide users with virtual machine that is easier to program underlying H/W
To manage the various resources of the system
User’s view of computer system depends to a great extent on the OS
OSs for distributed computing system broadly categorized into two types:
Network OS
Distributed OS
A distributed computing system that uses a network OS is usually referred to as a network system;
A distributed computing system that uses distributed OS is usually referred to as true distributed system or simply a distributed system