The optimal distribution of jobs among hosts in distributed environments is an important factor to achieve high performance. The optimal strategy depends on the application. In this paper we present a new online scheduling strategy for distributed EDI converter system. The strategy is based on the Bin-Stretching approach. The original algorithm has been enhanced to satisfy the business goals of meeting deadlines, priority processing, low response time and high throughput. The algorithm can be flexible adapted to different objective goals due to its two-stage strategy. We show by simulation and measurements on a real system that the modified Bin-Stretching approach fulfills the objective goals while requiring only low computational effort.
|Number of pages||11|
|Publication status||Published - Aug 2001|
- Load Balancing
- Parallel/Distributed Computing Systems
- Task Scheduling
- Performance Evaluation and Measurements