The Business Process Execution Language (BPEL) has become one of the predominant standards for Web services compositions oriented to business processes. An important capability of BPEL is that it allows processes to interact synchronously/asynchronously with other BPEL processes and/or choreographies at large-scale. However, it is a well-known fact that a greater system complexity leads to a greater probability of software/hardware failures. To attack the problem of failures for interactive business processes more efficient fault tolerance mechanisms are need. Nowadays there exist sophisticated solutions that tackle business processes fault tolerance, however, present drawbacks such as they affect the systems performance, they have a high implementation costs, they can jeopardize the scalability of the system and/or they do not support asynchronous BPEL communication among processes. In this paper, we propose to attack the dependability problem in a distributed manner, a fault tolerant solution based on communication-induced checkpointing (CiC) for interactive BPEL processes.
展开▼
