Power Based Performance and Capacity Estimation Models for Enterprise Information Systems

Historically, the performance and purchase price of enterprise information systems have been the key arguments in purchasing decisions. With rising energy costs and increasing power use due to the evergrowing demand for compute capacity (servers, storage, networks etc.), electricity bills have become a significant expense for todays data centers. In the very near future, energy efficiency is expected to be one of the key purchasing arguments. Having realized this trend, the Transaction Processing Performance Council has developed the TPC-Energy specification. It defines a standard methodology for measuring, reporting and fairly comparing power consumption of enterprise information systems. Wide industry adaption of TPC-Energy requires a large body of benchmark publications across multiple software and hardware platforms, which could take several years. Meanwhile, we believe that analytical power estimates based on nameplate power is a useful tool for estimating power consumption of TPC benchmark configurations as well as enterprise information systems. This paper presents enhancements to previously published energy estimation models based on the TPC-C and TPC-H benchmarks from the same authors and a new model, based on the TPC-E benchmark. The models can be applied to estimate power consumption of enterprise OLTP and Decision Support systems.