A long-running transaction model of workflow

Transaction of workflow systems composition has long-running feature which a global-transaction is divided into several sub-transactions. When the sub-transactions fail, it must back to the beginning of them, consistency property is preserved by using compensating transactions, which semantically undo the effects of the completed sub-transactions. However, the cost of compensation may be expensive and methods may be complex. To overcome this limitation, in this paper, a long-running transaction model named LTMBWF (Long-running transaction Model based on Workflow) is established. A novel scheduling algorithm named STCBCT (sub-transactions commit based on critical-transaction) is presented that and the resume algorithm is presented. The critical-transactions delay to commit as the cost of compensation is generally huge. Simulations show that STCBCT algorithm can confine the compensation sphere and reduce the cost of recovery.