Peak Power Analysis and Modelling

Database engines often consume significant power during query processing activities especially during complex query processing, which is motivating researchers to investigate the redesign of their database internals to minimize the energy overheads. While the prior literature has dealt exclusively with average power considerations, our focus here is on peak power consumption. We begin by profiling the peak power behavior of a representative suite of popular commercial database engines in benchmark environments, and demonstrate that their consumption can often be substantial and we have also shown that average power consumption behavior of optimizers are very different from the peak power consumption behavior of queries. Then, we develop a pipeline-based model of query execution plans that lends itself to accurately estimating peak power consumption, suggesting its gainful employment in server design and capacity planning. More potently, the model can be incorporated in current query optimizers to identify relatively “green” plans. We demonstrate sample instances of this application wherein power-hungry plans are replaced with alternatives that substantially reduce peak power requirements, without materially compromising query execution times. In the end we also show our preliminary work on inductive pipeline (i.e. when we have modeled a pipeline and a new pipeline comes which has just one join operation more so we can predict its peak power) and multiquery environment ( when more than one queries are running over a single server.