Dynamic Change Propagation for Process Choreography Instances

Business process collaborations realize value chains between different partners and can be implemented by so called process choreographies. Change has become a major driver for costly (re-)negotiations between the participants. Static a priori prediction models exist to calculate the feasibility of a change request prior to negotiation. However, the dynamic or behavioral aspect of choreography changes at the choreography instance level has not been investigated yet, i.e., the question whether a process choreography instance is compliant with the change request and hence allows for acceptance of the change request. This work takes the dynamic perspective and analyzes the impact of a single change request from one partner on the entire (distributed) choreography based on the notion of change regions and public check points. Change strategies are elaborated to ensure choreography instance state compliance. One transaction-based approach is specified using rollback regions. It identifies probabilistically the set of activity nodes to be compensated at all levels of the business collaboration to ensure state compliance. The technical evaluation enables observing the properties of the rollback region.