A Methodology for Generating Disk Drive Energy Models Using Performance Data

The ability to make intelligent decisions with respect to reducing the energy consumption of mechanical disk drives is dependent upon accurately tracking and accounting the runtime energy consumption of the mechanical and electrical components of the drive in both active and idle states. This work outlines a simple methodology for creating accurate hard drive runtime energy models through the use of easily obtainable data derived from published specifications and performance measurements. We use the generated models to create a TRADE estimator (TRAce-Driven disk drive Energy estimator). Validation of our estimator against a diverse range of disk drives shows that results obtained from the models are within 5% of measured results.