A Distributed Workflow Management System Model and Its Scheduling Algorithm

Workflow management system model and workflow task scheduling algorithm are the critical research areas in combination of workflow and grid techniques. On the basis of active network, a distributed, scalable workflow management system model which can adapt to dynamic changes of grid resource is presented. Workflow tasks are partitioned according to the characteristic of workflow structure, and the entire deadline and cost are divided into sub-deadline and sub-cost for each task partition based on its length and communication traffic; Considering user¿s multi-QoS requirements and communication time between workflow tasks, a trust and cost based Comprehensive Beneficial function and determination methods of trust¿s and cost¿s weights are provided. A scheduling algorithm TCD aims at comprehensive optimization of trust and cost is put forward and this algorithm tries to get an optimized solution for each task partition to achieve an optimized scheduling with multi-object for the entire workflow. Simulation results show that the algorithm can guarantee user¿s multi-QoS quite well.