Workflow scheduling in heterogeneous computing systems : A survey

Abstract

In the world of managing and processing large data sets significantly and executing such complex applications on heterogeneous systems, efficient scheduling is of prime concern. The workflow scheduling deals with mapping of tasks on the available resources while maintaining the precedence constraints and guaranteeing Quality-of-Service requirements and fairness in execution of workflow. Numerous workflow scheduling schemes have been presented by authors in the past. In this paper, we are providing the comprehensive survey of workflow scheduling techniques by classifying them based upon the workflow structures, scheduling criteria's, scheduling strategies in heterogeneous systems. We classified workflow-scheduling structures as DAG and Non-DAG structures, scheduling criteria's on the basis of performance driven, market driven and trust driven, workflow-scheduling strategies into Heuristic, Meta-Heuristic and Hybrid scheduling. We also considered the scheduling issues in various other fields of engineering; Hadoop Based, Map-Reduce Based, Agent Based. A comparative analysis of each scheduling algorithm is highlighted through their objectives, algorithm proposed, properties and environment in tabular form. The survey results are used to identify workflow scheduling approaches and the areas that need further research.