Neural network workflow scheduling for large scale Utility Management Systems

Abstract

Grid computing is the future computing paradigm for enterprise applications. It can be used for executing large scale workflow applications. This paper focuses on the workflow scheduling mechanism. Although there is much work on static scheduling approaches for workflow applications in parallel environments, little work has been done on a Grid environment for industrial systems. Utility Management Systems (UMS) are executing very large numbers of workflows with very high resource requirements. Unlike the grid approach for standard scientific workflows, UMS workflows have a different set of requirements and thereby optimization of resource usage has to be made in a different way. This paper proposes a novel scheduling architecture which dynamically executes a scheduling algorithm using near real-time feedback from the execution monitor. An Artificial Neural Network was used for workflow scheduling and performance tests show that significant improvement of overall execution time can be achieved by this soft-computing method.