Workflow Message Queue's Performance Effect Measurements on an EJB-Based Workflow Management System

摘要

The scalability and performance issues about large-scale workflow are the most impeccable topics, in recent, because the size of workflow system and its applications are becoming larger and larger and requiring, nevertheless, much more reliable and faster services. These technological pressures push us to develop an EJB-based workflow management system targeting at covering these two issues, which is named e-Chautauqua. In this paper, we present an asynchronous communication mechanism-workflow message queue mechanism-providing a reliable communication channel among the constituent objects organizing the enactment engine of e-Chautauqua, and try also to measure how much the mechanism effects on accomplishing the scalable and per-formable e-Chautauqua. Prom a series of test scenarios based on the message queue properties such as message alive time and maximum pool size, we have observed that the message alive time property sensitively effects on the scalability, and while on the other the maximum pool size property has a great influence on the performance of the workflow management system.