Two level demand response planning for retail multi-tenant datacenters.
Resumen:
This article presents a two level planning approach for the participation of datacenters in the electricity market, by providing appropriate actions to demand response events, a relevant issue to contribute to the smart grid paradigm. A planning strategy is proposed for retail collocation datacenters, considering their heterogeneity. A negotiation is applied at the datacenter level and scheduling heuristics are used at the tenants level. Three heuristics are evaluated, accounting for different features of tasks submitted for execution. The proposed approach is evaluated considering realistic infrastructures (servers and air conditioning system) and workloads. The main results suggest that the proposed approach provides appropriate plannings for demand response events, significantly improving over Business-as-Usual operation and when there is not electricity load shedding. The Nash+PL heuristic computed the best results regarding cost while accounting for appropriate quality of service.
2021 | |
Agencia Nacional de Investigación e Innovación FSE_1_2017_1_144789 | |
Demand response Computational intelligence Planning, Datacenters |
|
Inglés | |
Universidad de la República | |
COLIBRI | |
https://hdl.handle.net/20.500.12008/26941 | |
Acceso abierto | |
Licencia Creative Commons Atribución - No Comercial - Sin Derivadas (CC - By-NC-ND 4.0) |