Scheduling computational and transmission tasks in computational grids

Abstract

Computational grid is a computing environment dedicated to execute applications with large computational requirements. These applications are mainly scientific applications developed for using in many scientific areas. One type of such applications are workflow application which consists of a set of precedence constrained computational tasks. The precedence constraints are caused by transmission of data files and/or the control flow between two computational tasks. Three models of the problem of resource allocation and scheduling workflow application in the computational grid are considered. They differ among themselves in assumptions about workflow applications and computer network. For the first model with distributed resources we present how to adopt metaheuristics developed for MRCPSP. For the second model with setup operations we use the idea of schedule-dependent setup operations and also present the adaptation of a metaheuristic. For the third model with transportation network we present how to find a feasible resource allocation.