An adaptive service for grid applications

Abstract

The grid is a dynamic system where resources are subject to change due to system performance degradation, system failure, etc. The task of providing commitments and assurances for end-users' jobs is non-trivial, since a grid system can integrate heterogeneous resources with varying quality and availability. This places importance on the system's ability to monitor the state of these resources and adapt to changes in their availability over time. We propose an adaptive service based on a reflective technique that supports job migration to ensure timely job completion. A key feature of our approach is that the user is not required to install additional software, or make complex alterations to their code requiring specialist grid computing knowledge. We evaluate our approach using an industrial application on the White Rose grid.

Keywords:

• grid computing
• adaptive services
• reflection
• checkpointing
• migration

Acknowledgements

The work reported in this paper was developed and undertaken as part of the BROADEN project. The BROADEN project is part-funded via the UK Department of Trade and Industry under its Technology Innovation Programme. The work described was undertaken by teams at the Universities of Leeds, York and Sheffield with industrial partners Rolls-Royce, EDS, Oxford BioSignals and Cybula Ltd.