| In contrast to the traditional central workflow management system, the distributed workflow management system has not only the function of the traditional central workflow management, but also the advantages such as the high expansibility, high efficiency, high fault-tolerant and high dependability. The distributed workflow engine is the core of the distributed workflow management system, the design of it will impact the performance of the workflow management system greatly, so how to design the high-performance distributed workflow engine is the content of this paper studies.By studying and analyzing the traditional workflow models, this paper summarizes the deficiencies and disadvantages of the traditional workflow patterns. Aiming at the particularities of the distributed workflow management system and the requirements of the transaction processing according to the enterprise's applications, this paper puts forward a high performance distributed workflow engine model. Comparing to other distributed workflow engines, this workflow engine has better high dependability, better high fault-tolerant and better transaction processing mechanism. In order to realizing the three goals above, this paper first introduces the following three conceptions into distributed workflow engine, and goes on studying and analysis deeply, and also give the implementation methods of them: (1) Splits the dispatch of the workflow engine into workflow process dispatch and task dispatch, so makes task scheduling to separate out from t he s cheduling o f t he d istributed workflow engine, in order toimprove the reliability and the throughput of the workflow system applies the two-phase scheduling method to schedule them. (2) In order to improve the fault-tolerant, brings forward a conception named the workflow instances redundancy concurrence. (3) Applying the transaction management thought based on the web service to fulfill the requirements of the transaction processing of the workflow management dealing.
|
PDF Full Text Request