COORDINATING PARALLEL PROCESSES ON NETWORKS OF WORKSTATIONS

The network of workstations (NOW) we consider for scheduling is hetero geneous and nondedicated, where computing power varies among the works tations and local and parallel jobs may interact with each other in ex ecution. An effective NOW scheduling scheme needs sufficient informati on about system heterogeneity and job interactions. We use the measure d power weight of each workstation to quantify the differences of comp uting capability in the system. Without a processing power usage agree ment between parallel jobs and local user jobs in a workstation, job i nteractions are unpredictable, and performance of either type of jobs may not be guaranteed. Using the quantified and deterministic system i nformation, we design a scheduling scheme called self-coordinated loca l scheduling on a heterogeneous NOW. Based on a power usage agreement between local and parallel jobs, this scheme coordinates parallel proc esses independently in each workstation based on the coscheduling prin ciple, We discuss its implementation on Unix System V Release 4 (SVR4) . Our simulation results on a heterogeneous NOW show the effectiveness of the self-coordinated local scheduling scheme. (C) 1997 Academic Pr ess.