A reliable framework for adaptive scientific workflow management systems based on SOA

In this paper a reliable framework for scientific workflow management system is presented. The reliability of workflow management systems could be guaranteed by supplying workflow engines with fault control components. A fault control component may receive fault notifications through exceptions raised by web service handlers. Receiving a fault exception, the program execution state has to be rolled back to the state before the fault. To achieve this, the execution state of each component could be kept in a file, separately. Using the log kept in this file, the program status could be returned to the states before invoking the fault web service. The program execution is continued after the fault web service is substituted with a correct one. The applicability of the proposed framework is practically evaluated by applying it to design three different scientific workflow engines.