Coordinating large distributed relational process structures

Abstract

Representing a business process as a collaboration of interacting processes has become feasible with the emergence of data-centric business process management paradigms. Usually, these interacting processes have relations and, thereby, form a complex relational process structure. The interactions of processes within this relational process structure need to be coordinated to arrive at a meaningful overall business goal. However, relational process structures may become arbitrarily large. With the use of cloud technology, they may additionally be distributed over multiple nodes, allowing for scalability. Coordination processes have been proposed to coordinate relational process structures, where processes may have one-to-many and many-to-many relations at run-time. This paper shows how multiple coordination processes can be used in a decentralized fashion to more efficiently coordinate large, distributed process structures. The main challenge of using multiple coordination processes is to effectively realize the coordination responsibility of each coordination process. Key components of the solution are the subsidiary principle and the hierarchy of the relational process structure. Finally, an implementation of the coordination process concept based on microservices was developed, which allows for fast and concurrent enactment of multiple, decentralized coordination processes in large, distributed process structures.

Remote Service Manager using Zookeeper

“…In the paper [20] coordination mechanisms for managing large distributed relational process structures are introduced by the object-aware process management method, which emphasizes decentralization and distribution to boost performance and maintainability. The process consists of phases for decentralized coordination, a switch from central to decentralized coordination, and a validation of the proof-ofconcept.…”