Predictive resource management for meta-applications

This paper defines meta-applications as large, related collections of compu tational tasks, designed to achieve a specific overall result, running on a (possibly geographically) distributed, non-dedicated mete-computing platfo rm. To carry out such applications in an industrial context, one requires r esource management and job scheduling facilities (including capacity planni ng), to ensure that the application is feasible using the available resourc es,that each component job will be sent to an appropriate resource, and tha t everything will finish before the computing resources are needed for othe r purposes. This requirement has been addressed by the PAC in three major European coll aborative projects: PROMENVIR, TOOL-SHED and HPC-VAO, leading to the creati on of job scheduling software, in which scheduling is brought together with performance modelling of applications and systems, to provide meta-applica tions management facilities. This software is described, focusing on the pe rformance modelling approach which was needed to support it. Early results from this approach are discussed, raising some new issues in performance modelling and software deployment for meta-applications. An ind ication is given about ongoing work at the PAC designed to overcome current limitations and address these outstanding issues. (C) 1999 Published by El sevier Science B.V. All rights reserved.