Grid Workflow Acceleration using a Dataflow Paradigm

摘要

Data flow is one of the radical approaches to Grid Computing. This paper introduces the leading edge approach of system performance acceleration through Grid Workflow Acceleration by Dataflow Paradigm. To simulate the aforementioned topic, an artificial environment is created, where three tasks need to accomplish using all available resources. Introducing dataflow to workflow has immerged as possible solution to improve the performance of computation of this issue. This paper solves the job using traditional workflow strategy and then it solved using the proposed method and compares the performance. The instance is tested on Parallel Computing Tool of MATLAB, which allows programs to be executed in a cluster of computer environment. The system generates jobs randomly and assuming that the complexity of a particular job, forces the system to partition the job. It then processes the partitions independently, and later reassembles the results into one output stream. The significant performance improvement of this dataflow based workflow system is a convincing outcome for future grid.