The project explores system-level energy management schemes for storage systems. It aims to develop storage organizations, groupings, and data layouts that facilitate energy conservation by exploiting complex redundancy already built into the system for fault tolerance. The approach is motivated by the trend towards using increasingly sophisticated redundancy techniques in storage systems to guard against data loss. The project investigates methods to leverage this redundancy to provide additional functionality, in particular energy conservation. The project involves several components: exploration of the design space using replication as well as erasure-coding techniques for energy conservation, design of control mechanisms and algorithms for dynamic disk provisioning and associated performance and energy models, and development of a test-bed for empirical validation of the techniques.
Energy consumption has become an increasingly important issue in the operations of large-scale storage systems. Power and cooling equipment, and electricity together represent a significant portion of the total cost of ownership. The results of this project have potential for economic and environmental impact on the operation of large storage centers.
