Optimizing benchmark configurations for energy efficiency (abstracts only)

Historically compute server performance has been the most important pillar in the evaluation of datacenter efficiency, which can be measured using a variety of industry standard benchmarks. With the introduction of industry standard servers, price-performance became the second pillar in the 'efficiency equation'. Today with an increased awareness in the industry for power optimized designs and corporate initiatives to reduce carbon emissions, data center efficiency needs to incorporate yet another key element in this equation: energy efficiency. Initial models based on 'name-plate' power consumption have been used to estimate energy efficiency while recently industry standard consortia like SPEC, TPC and SPC have started amalgating new energy metrics with their traditional performance metrics. TPC-Energy, enables the measuring and reporting of energy efficiency for transaction processing systems and decision support systems. In this paper we analyze TPC C benchmark configurations that may achieve leadership results in TPC-Energy using existing, more energy efficient technologies, such as solid states drives for storage subsystems, low power processors and high density DRAM in back end server and middle tier systems. Even though the study is based on TPC-C configurations these configuration optimizations are applicable to other benchmarks and production systems alike. We envision that the energy efficiency metrics and related optimizations to claim benchmark leadership will accelerate development and qualifications of energy efficient component and solutions.