Open Source Workflow: A Viable Direction for BPM?

With the growing interest in open source software in general and business process management and workflow systems in particular, it is worthwhile investigating the state of open source workflow management. The plethora of these offerings (recent surveys such as [4, 6], each contain more than 30 such systems) triggers the following two obvious questions: (1) how do these systems compare to each other; and (2) how do they compare to their commercial counterparts. To answer these questions we have undertaken a detailed analysis of three of the most widely used open source workflow management systems [1]: jBPM4, OpenWFE5, and Enhydra Shark6. Another obvious candidate would have been the open-source workflow management system YAWL (www.yawlfoundation.org). However, given the the authors’ close involvement in the development of YAWL, we did not include it in our evaluation. This analysis was based on the workflow patterns framework [2]. This framework provides a collection of generic constructs which recur in a workflow context. It is divided into control-flow, data, and resource patterns based on the process perspectives outlined in [3]. A patterns-based analysis is guided by explicit evaluation criteria which are identified for each pattern. It aims to investigate the ability of a workflow system to support each of the patterns that have been identified and is based on the premise that each pattern describes a feature that it is desirable to support in a business process context. Hence, the workflow patterns framework is not concerned with expressive power, but rather with suitability (see e.g. [5]). We choose to use the workflow patterns as the basis for our investigation because it is a well established framework that is widely used for WFMS evaluations (as evidenced by the numerous references to it). There are already a substantial number of evaluations of contemporary offerings based on the patterns and they provide an effective means of comparing the capabilities of differing systems on a neutral basis. For the purposes of this analysis, the results from some of these earlier evaluations (i.e., Staffware, WebSphere MQ and Oracle BPEL PM) are added to the results from the analysis of open source systems summarized here.