Parallel memetic algorithms for independent job scheduling in computational grids
In this chapter we present parallel implementations of Memetic Algorithms (MAs) for the problem of scheduling independent jobs in computational grids. The problem of scheduling in computational grids is known for its high demanding computational time. In this work we exploit the intrinsic parallel n...
| Autores: | , |
|---|---|
| Tipo de recurso: | capítulo de libro |
| Fecha de publicación: | 2008 |
| País: | España |
| Institución: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/128958 |
| Acceso en línea: | https://hdl.handle.net/2117/128958 https://dx.doi.org/10.1007/978-3-540-70807-0 |
| Access Level: | acceso abierto |
| Palabra clave: | Computational grids (Computer systems) Memetic algorithms Independent job scheduling Parallelisation Two level granulation Computació distribuïda Àrees temàtiques de la UPC::Informàtica::Informàtica teòrica::Algorísmica i teoria de la complexitat |
| Sumario: | In this chapter we present parallel implementations of Memetic Algorithms (MAs) for the problem of scheduling independent jobs in computational grids. The problem of scheduling in computational grids is known for its high demanding computational time. In this work we exploit the intrinsic parallel nature of MAs as well as the fact that computational grids offer large amount of resources, a part of which could be used to compute the efficient allocation of jobs to grid resources. The parallel models exploited in this work for MAs include both fine-grained and coarse-grained parallelization and their hybridization. The resulting schedulers have been tested through different grid scenarios generated by a grid simulator to match different possible configurations of computational grids in terms of size (number of jobs and resources) and computational characteristics of resources. All in all, the result of this work showed that Parallel MAs are very good alternatives in order to match different performance requirement on fast scheduling of jobs to grid resources. |
|---|