A Dataflow Optimization Mechanism for Service-Oriented Cloud Workflow

While service-oriented cloud workflow shows potentials of inherent scalability, loose coupling and expenditure reduction, such issues as data transfer and efficiency have popped up as major concerns. In this paper, a dataflow optimisation mechanism is proposed, including data transfer strategy and data placement strategy. The data transfer strategy uses integrated data structures to enable the elimination of the SOAP serialisation and de-serialisation within the web service invocation by the exchange of their references. In addition, the data placement strategy groups the existing datasets and dynamically clusters newly generated datasets to minimise data transfers while keeping a polynomial time complexity. By integrating the optimisation mechanism into service-oriented cloud workflow system, we can expect efficiency improvements in data transfer and workflow execution. Experiments and analysis supported our efforts.