Integrated Data and Process Management: Finally?

Contemporary information systems are generally built on the principle of segregation of data and processes. Data are modeled in terms of entities and relationships while processes are modeled as chains of events and activities. This situation engenders an impedance mismatch between the process layer, the business logic layer and the data layer. We discuss some of the issues that this impedance mismatch raises and analyze how and to what extent these issues are addressed by emerging artifact-centric process management paradigms. 1 The Data Versus Process Divide Data management and process management are both well-trodden fields – but each in its own way. Well-established data analysis and design methods allow data analysts to identify and to capture domain entities and to refine these domain entities down to the level of database schemas in a seamless and largely standardized manner. Concomitantly, database systems and associated middleware enable the development of robust and scalable data-driven applications, while contemporary packaged enterprise systems support hundreds of business activities on top of shared databases. In a similar vein, well-documented and proven process analysis and design methods allow process analysts to identify and to capture process models at different levels of abstraction, ranging from high-level process models suitable for qualitative analysis and organizational redesign down to the level of executable processes that can be deployed in Business Process Management Systems (BPMS). But while data management and process management are each well supported by their own body of mature methods and tools, these methods and tools are at best loosely integrated. For example, when it comes to accessing data, BPMS typically rely on request-response interactions with database applications or packaged enterprise systems. Typically, data fetched from these systems are copied into the “working memory” of the BPMS. The data in this working memory are then used to evaluate business rules relevant to the execution of the process, and to orchestrate both manual and automated work. But the burden of synchronizing the working data maintained by the BPMS with the data maintained by the underlying systems is generally left with the developers. More generally, the “data vs. process” divide leads to an impedance mismatch between the data layer, the business logic layers and the process layer, which in the long run, hinders on the coherence and maintainability of information systems. In particular, the data vs. process divide has the following effects: – Process-related and function-related data redundancy. The BPMS maintains data about the state of the process, since these data are needed in order to enable the