Design Issues of Distributed Computing

various resources of a computersystem must be protected againstdestruction and unauthorized access.

Enforcing security in a distributed system ismore difficult because of the lack of asingle point of control and the use ofinsecure networks for data communication

manages multiple resources, must bedesigned properly to increase the system'sreliability

higher reliability needed to avoid faults,tolerate faults and to detect andrecover from faults

Commonly used methods are faultavoidance and fault tolerance.

Determines whether the system can beextended in various ways withoutdisrupting existing system and services

should be flexible due: Ease ofmodification & Ease of enhancement

heterogeneous distributed system consistsof interconnected sets of dissimilarhardware or software systems

diversity causing heterogenousdistributed system far more difficult formhomogenous

heterogenous preferred as it provideflexibility

overall performance should be better thanor at least equal to that of running thesame application on a singleprocessorsystem

design principle should be consideredbatch, cache, minimize copying of data,minimize network traffic, take advantage offinegrain parallelism for multiprocessing

essential part of distributed systems e.g.,clients and servers must communicate forrequest and response

Synchronous or blocking senderwaits for receiver to execute a receiveoperation

Asynchronous or nonblocking

no common clock among the multipleprocessors

difficult task and requires that severaldifferent aspects of transparency

messages could even be lost

delivery of messages is delayed