TMSS: a task management and scheduler system in cluster for remote computing service

Abstract

Cluster is an aggregation of computers to obtain powerful computing energy. In this paper, we discuss the ability of its power to provide remote service. We focus on the convenience for remote users to submit their tasks, providing a friendly environment to run the tasks efficiently, and maintaining the equity of the service between each user. To achieve this outcome, the scheduling method must avoid the properties that will benefit the system performance but impair the equity between users as well. We also supply users with limited means to enhance their task running speed and executive environment. These means exert a considerable influence on the system performance, but at the same time expend more than the average approach. The variance is embodied in the final evaluation of the service quality the user obtains. With the above character, the system presents a little differently than traditional ones, which consider mostly the cluster load balance and performance. We also regard the system balance as an important standard. However, this concern is based on sustaining a credible service.