Executing jobs with deadline constraints in a privately owned workstation environment

Abstract

A system that considers users' deadlines when scheduling background jobs is described which will enhance the quality of the computing environments it serves by exploiting underutilized workstation capacity without interfering with an owner's local workstation stage. By developing and evaluating a simulation model of the structure, the author shows that this scheduler can effectively schedule jobs according to their deadline constraints to improve the quality of service for the users. In the proposed approach, adaptive techniques for estimating the amount of capacity available for sharing in a workstation cluster are combined with user supplied information of the expected demand of their jobs. The improvement in deadline miss ratio due to the proposed structure was evaluated by means of simulation. The scheduling structure showed improvements over a wide range of user loads and tightness of deadlines.<>