Multi criteria biased randomized method for resource allocation in distributed systems: Application in a volunteer computing system
Volunteer computing is a type of distributed computing in which a part or all the resources (processing power and storage) necessary to run the system are donated by users. In other words, participants contribute their idle computing resources to help running the system. Due to the fact that the nod...
| Autores: | , , , |
|---|---|
| Formato: | artículo |
| Estado: | Versión aceptada para publicación |
| Fecha de publicación: | 2018 |
| País: | España |
| Recursos: | Varias* (Consorci de Biblioteques Universitáries de Catalunya, Centre de Serveis Científics i Acadèmics de Catalunya) |
| Repositorio: | Recercat. Dipósit de la Recerca de Catalunya |
| OAI Identifier: | oai:recercat.cat:10230/59084 |
| Acesso em linha: | http://hdl.handle.net/10230/59084 http://dx.doi.org/10.1016/j.future.2017.11.039 |
| Access Level: | acceso abierto |
| Palavra-chave: | Distributed computing Volunteer systems User assignment Allocation methods Resource provisioning |
| Resumo: | Volunteer computing is a type of distributed computing in which a part or all the resources (processing power and storage) necessary to run the system are donated by users. In other words, participants contribute their idle computing resources to help running the system. Due to the fact that the nodes which compose the system are provided by a large number of users instead of a single (or a few) institution, a main drawback of volunteer computing is the unreliability of these nodes. For this reason, the selection of nodes to be involved in each task becomes a key issue. In this paper, we propose the Multi Criteria Biased Randomized (MCBR) method, a novel selection method for large-scale systems that use unreliable nodes. MCBR method is based on a multicriteria optimization strategy. We evaluated the method in a microblogging social network formed by a large number of microservices hosted in nodes voluntarily contributed by their participants. Simulation results show that our proposal is able to select nodes in a fast and efficient manner while requiring low computational power. |
|---|