Adapting Predictions and Workloads for Power Management

Power conservation in systems is critical for mobile, sensor network, and other power-constrained environments. While disk spin-down policies can contribute greatly to reducing the power consumption of the storage subsystem, the reshaping of the access workload can actively increase such energy savings. Traditionally reshaping of the access workload is a result of caches passively modifying the workload with the aim of increasing hit ratios and reducing access latency. In contrast, we present the a shifting predictive policy that actively reshapes the workload with the primary goal of conserving disk energy consumption. By reshaping the disk workload to explicitly lengthen idle periods, the disk can remain spun-down longer, saving more energy. We show that our approach can save up to 75% of disk energy compared to the common fixed-timeout spin-down policies. Our shifting algorithm dynamically shifts to the most energy efficient cache prefetching policy based on the current workload. This best shifting prefetching policy is shown to use 15% to 35% less energy than traditional disk spin-down strategies and 5% to 10% less energy than the use of a fixed (non-shifting) prefetching policy.