Design And Implementation Of Grid Job Management Service

As an important module within grid system software, the grid job management service is responsible for shielding the heterogeneous and autonomous features of grid computing resources, managing the life cycle of grid jobs, as well as providing standard and easy-to-use user interfaces for grid users who need job-related functions. It's necessary to design and implement a grid job management service which ensures plenty functions and high performance.In this paper, we firstly introduce the background and related works about grid job management service, including a functional and performance analysis on the most popular grid job management services nowadays, an introduction on the grid job scenarios and user patterns, as well as a review on the related global standards.Then we propose our research goal: to design and implement a grid job management service for VEGA GOS, which ensures a rich set of functions and high performances.To accomplish this goal, we analysis the key problems involved in designing and implementing grid job management services, including it's deploying scenario and user requirements. And then present our VegaBatch grid job management service based on VEGA GOS.VegaBatch is able to provide plenty of useful job-managing functions, including querying the global job records as well as rerunning the grid jobs. We also enhance the performance of VegaBatch via several mechanisms such as the meta-info cache mechanism, and the abstract-zipping mechanism.Finally, we describe some experiments on VegaBatch with other popular grid job management services. The result shows that, VegaBatch is able to provide a high-performance mechanism for managing grid jobs, which ensures the value of VEGA GOS platform, and fulfills the job-related requirement of Grid users.