Study Of Anchorage Server Fault-Tolerance In Migrating Workflow System

Migrating workflow is a new technology which applies mobile computing to workflow management.Workflow business process is mapped into one or more migrating instances according to the compelexity of its business goal, every migrating instance executes a sub-process of relatively independent goal,and it implements its total business goal by migrating and cooperating with the support of workplace.The three essensial factors of migrating workflow management system include migrating work management engine,migrating instace and workplace.The workflow management engine defines workflow process,creates migrating instace and coordinates multi-instaces.Workplace includes anchorage server and assiocated work-machine network.The anchorage server is the place where migrating instance run,while the work-machine is the actual place where workflow service gets executed.The migrating instance would inevitably be affected by many inrelibale factors in the environment where complex network,heteromerous systems exist.The crash of anchorage server host would lose all the migrating instances on it,the fault of running environment for migrating instance and local workflow service would cause business process to block.The fault-tolerance performance has become a bottle-neck for the wide adoption of migrating workflow system.Fault-tolerance is one of the most important research content in the field of migrating workflow management system.The mechanism of fault-tolerance on workplace is dicussed in this article.Based on the reseach of the activity of migrating instance and the feature of workflow management,the mechanism should have two basic properties:non-block,executed-only-once.The total-duplicated primary-backup schema based on software is propused to achive non-block property,while to achive executed-only-once,the definition of business checkpoint is propused,which acts as the atomic transaction.Three circumstances including fault-tolerance of anchorage server host,of runnig environment and workflow service has been discussed in this article,and related policies for detecting and recovering has been made.A fault-tolerance component has been realized on the kernel of anchorage server according to the fault-tolerance model,which has high extension besides good performance.The result of testing shows that the prototype successfully realize its original goal,prove the accuracy of the fault-tolerance model.