期刊名称：International Journal of Database Management Systems
印刷版ISSN：0975-5985
电子版ISSN：0975-5705
出版年度：2017
卷号：9
期号：1
页码：1
DOI：10.5121/ijdms.2017.9101
出版社：Academy & Industry Research Collaboration Center (AIRCC)
摘要：Energy consumption has become a first-class optimization goal in design and implementation of dataintensivecomputing systems. This is particularly true in the design of database management systems(DBMS), which was found to be the major consumer of energy in the software stack of modern datacenters. Among all database components, the storage system is one of the most power-hungry elements. Inprevious work, dynamic power management (DPM) techniques that make real-time decisions to transitionthe disks to low-power modes are normally used to save energy in storage systems. In this paper, we tacklethe limitations of DPM proposals in previous contributions. We introduced a DPM optimization modelintegrated with model predictive control (MPC) strategy to minimize power consumption of the disk-basedstorage system while satisfying given performance requirements. It dynamically determines the state ofdisks and plans for inter-disk data fragment migration to achieve desirable balance between powerconsumption and query response time. Via analyzing our optimization model to identify structuralproperties of optimal solutions, we propose a fast-solution heuristic DPM algorithm that can be integratedin large-scale disk storage systems for efficient state configuration and data migration. We evaluate ourproposed ideas by running simulations using extensive set of synthetic workloads based on popular TPCbenchmarks. Our results show that our solution significantly outperforms the best existing algorithm inboth energy savings and response time.
关键词：Database Management System; Dynamic Power Management; Model Predictive Control
