Improving Performance of Composite Web Services

Composite Web Services (CWS) aggregate multiple Web Services in one logical unit in order to accomplish a complex task (e.g. business process). This orchestration is typically achieved by use of a workflow language. Workflows facilitate the process of aggregating existing atomic and other CWS into new service layers. However due to numerous consumers and possible fluctuations in their arrivals the services performance under various loads becomes an important issue. Service compositions exposed to transient overloads expose problematic behaviour due to complex interactions of the underlying services. This in its turn usually results in the performance degradation. This paper proposes employing scheduling service requests in order to improve the overall CWS performance in overload situations. Different scheduling policies are evaluated for the CWS workflow patterns sequence and split-synchronization. In addition the paper presents scheduling policy called Augmented Least Work Reaming (ALWKR), that extends LWKR by taking advantage of existing workflow topology information.