Static allocation of communicating processes for distributed computing systems with resource heterogeneity

Abstract

Allocation of processes to the processing nodcs in such a way as to minimise interprocossor communication and and, also ensure load balancing is a major problem in distributed computing system. The problem is even more complicated if the system is heterogeneity. In this paper we have presented the affinity graph model for performing process allocation for distributed computing system and augment the model to cover distributed systems with heterogeneity of resourccs. The main  advantages of the affinity graph model are its ability to reflect both the load balancing as well as the minimisatmn of interprocessor communication criteria in a single representation and the facility for varying the weightage given to tht above two criteria. We have presented on algorithm for, performing allocation for a two - processor system and also briefly illustrate how allocation can be done for systems with a large number of processors by using a binary-tree-structured system as an example.