Economic Models for Allocating Resources in Computer Systems

With the advances in computer and networking technology thousands of heterogeneous com puters can be interconnected to provide a large collection of computing and communication resources These systems are used by a growing and increasingly heterogeneous set of users A macroscopic view of distributed computer systems reveals the complexity of the organi zation and management of the resources and services they provide This complexity arises from size e g no of systems no of users and heterogeneity in applications e g on line transaction processing multimedia intelligent information search and resources CPU memory bandwidth locks naming services The complexity of resource allocation is further increased by several factors First in many distributed systems the resources are in fact owned by multiple organizations Second the satisfaction of users and the performance of applications is determined by the simultane ous allocation of multiple resources A multimedia server application requires I O bandwidth to retrieve content CPU time to execute server logic and communication protocols and net working bandwidth to deliver the content to clients The performance of applications may also be altered by trading one resource for another For example the multimedia server ap plication may perform better by releasing memory and acquiring higher CPU priority This trade may result in smaller bu ers for I O and networking but improve the performance