Parallel processing DNA sequences on multicluster and grid architectures : Software overhead

A DNA sequence analysis parallelization in large databases using cluster, multi-cluster, and GRID is presented. Achievable speedup, scalability, and overhead introduced by communications are discussed, and the impact of the Grid middleware on the performance obtained with clusters is detailed. The experimental work carried out with homogeneous and heterogeneous clusters is presented, along with a comparison of the results obtained when migrating the algorithms to a GRID. Finally, current lines of work related to the study of models and paradigms for the resolution of parallel algorithms on GRID architectures are presented. Workshop de Procesamiento Distribuido y Paralelo (WPDP) Red de Universidades con Carreras en Informática (RedUNCI)