Queuing analysis and performance evaluation of workflow through WFQN

Performance prediction is one of the most important research topics of workflow. To investigate the performance of workflow systems in queuing condition, this paper extends traditional WF-net into WFQN (WF queuing network), by modeling tasks as FIFS (first-in-first-service) queues and the source place as the input of tokens following poisson arrival process. Analytical methods are introduced to evaluate the queue-length, wait-time and completion-duration. The case study (especially the case of airline ticket booking application) shows that WFQN can model real-world workflow-based applications effectively. Through Montecarlo simulations in the case study, we show analytical models are verified by simulative results. We also present a sensitivity analysis technique to identify performance bottle-necks of WFQN. This paper concludes with a comparison with relate work.