Distributed Workflow Execution Optimization of SBA Based on Service-Oriented Architecture

Abstract

According to the characteristics of the information resources in modern acquisition, such as heterogeneous, distributed, loosely coupled, the problems and deficiencies of workflow technology and products were analyzed. Dynamic service grouping strategies based on SOA was proposed and was proved in theory. In addition, the workflow architecture based on SOA was built and the workflow engine of simulation-based acquisition (SBA) was designed. Therefore, the legacy code can be wrapped and the service can be grouped dynamically from the dynamic generic service factory. Grouping strategy was proved according to the SPOT satellite image application which was part of the SBA toolbox on grid. Experimental results show that: services grouping of a workflow can reduce the total overhead induced by the submission, scheduling, queuing and data transfer time. On a production grid infrastructure, the optimization proposed introduces a significant speed-up when compared to a traditional execution. The range is from 1.21 to 2.97.