Scheduling Different Types of Bag-of-Tasks Jobs in Distributed Systems

Abstract

When distributed processors are shared by bag-of-tasks applications, we are facing the problem of efficient assignment of processors to tasks and of efficient task scheduling in each processor. This paper studies scheduling of different types of bags-of-tasks in a distributed system where two algorithms of task routing are considered in combination with two scheduling techniques. The aim is to study the performance of the scheduling techniques in each task routing algorithm case, taking also into account fairness in task execution order. A simulation model is used to examine the performance of the scheduling strategies in different cases of sporadic job arrivals. The results obtained from the simulation experiments show that the routing policies play a more important role in performance than the scheduling of tasks in their assigned queues. It is also shown that the performance depends on the arrival process and the type of bag-of-tasks applications.