Semi-Static Algorithms for Data Replication and Scheduling Over the Grid

Abstract — Managing large datasets has become one major application of grids. Life science applications usually manage large databases that should be replicated to let applications scale. The growing number of users and the simple access to Internet-based applications has stressed grid middlewares. Such environments are thus asked to manage data and schedule computation tasks at the same time. These two important operations have to be tightly coupled to get the best results. This paper presents several heuristics that combines data management and scheduling using a steady-state approach. Starting from an intial distribution given by the resolution of a linear program, we optimize the results at run-time to improve the quality of the data mapping and the scheduling of computation requests when the request distribution evolves. Our theoretical results are validated using simulation and logs from a large life science application. I