A collaborative processes synchronization method with regards to system crashes and network failures.

Processes can synchronize their states by exchanging messages. System crashes and network failures may cause message loss, so that state changes of a process may remain unnoticed by its partner processes, resulting in state inconsistency or deadlocks. In this paper we define a method to transform a business process into its recovery-enabled counterpart. We also discuss the correctness proof of the transformation, and the performance evaluation of our prototype implementation. In our previous work, we presented solutions to these synchronization problems that were based on rather strong assumptions. For example, specific failure patterns or interaction patterns (one client instance interacts with one server instance) were assumed. In this paper, the solution is extended to multiple process instances with more possible synchronization failures.