An Autonomic Performance-Aware Workflow Job Management for Service-Oriented Computing

Workflow job is composed of several ordered subtasks which invoke different computing services. These computing services are deployed on geographically distributed servers. Towards Workflow Job Management, how to schedule workflow jobs to achieve high server resource utilization and how to ensure Quality of Service (QoS) pose several challenges. In the paper, an autonomic Performance-Aware Workflow Job Management is proposed. It firstly decomposes workflow jobs into subtasks, and then adopts Virtual Allocation Strategy, which utilizes the concept of virtual queue, to allocate them to the servers. We also apply a Detection Adjustment Approach for Virtual Allocation to dynamically adjust workload of each computing server according to the real-time system workload changes. Additionally, it also utilizes Occupy Allocation to ensure the QoS. These capacities enable our Workflow Job Management adaptable and autonomic. Finally we establish simulations to demonstrate system performance and QoS.