Abstract:
This work shows that the use of a temporal Database Management System (DBMS) can be advantageous in Business Process Management (BPM). Compared to the use of a conventional DBMS, the support of a temporal query language and the respective temporal query processing is a major advantage. It enables a less difficult retrieval of process execution data. Additionally, query planning and execution times have been evaluated. In some cases, also a better execution performance could be observed in the temporal database. However, in this respect no general advantage could be identified. To evaluate the execution performance of both DBMS, thirteen BPM-relevant query statements have been implemented in standard and temporal SQL and executed in a standard and a temporal PostgreSQL installation. Second is a PostgreSQL extension developed as research prototype by Dignös, Böhlen, Gamper and Jensen and described in their work Extending the Kernel of a Relational DBMS with Comprehensive Support for Sequenced Temporal Queries. In order to enable a comprehensive comparison of the query execution performance, three differently sized sets of process execution data were generated within a simulation application, utilizing the Activiti workflow engine and its underlying data model. Up to 4,000 process iterations of a sample credit approval process have been simulated. The generated data has been queried repeatedly utilizing standard and temporal SQL statements being executed in the respective DBMS.
