Research Of Workflow Scheduling With Communication Overhead

Tasks’ data communication is ignored in the most of workflow scheduling model. In the cloud computing, users submit workflow to multiple resources to executive, there is a large amount of data between interactions. And the network bandwidth is not unlimited in the cloud networks. Tasks’ data communication means the negligible communication overhead, and it should be added to the model. So, this paper focuses on time optimization with cost-constrained and workflow multi-dimensional optimization problems and propose communication-aware workflow time optimization model and time-cost optimization model.In the communication-aware workflow time optimization model, propose a critical service algorithm and two communication-aware iterative rules. Critical service algorithm chooses the critical service for every task to get the minimum finish time of workflow. By incorporating communication overhead with priority rule MPBL and MPTL, two new communication-aware priority rules, i.e. CAMPBL and CAMPTL are designed. Both iterative rules aim to increase minimum time for maximum reduction of the costs, optimize costs gradually, until meet budget-constraints.In the communication-aware workflow time-cost optimization model, improve the previous BLMax algorithm and BLMin algorithm, rebuild a new criteria with communication overhead and load factor, propose two new algorithm, including CABLMax algorithm and CABLMin algorithm, get better time and cost according the metrics computed by users’ scheduling target in the new scheduling model.Build new models and achieve the MPBL, MPTL, CAMPBL, CAMPTL rule in the communication-aware workflow time optimization model, and BLMax, BLMin, CABLMax, CABLMin algorithm in the communication-aware workflow time-cost optimization model by My Eclipse, and test the performance and efficiency with different workflow instances. Experimental results show that, in the new communication-aware scheduling model, the new scheduling algorithms can optimize the time and cost.