Architectural Implications of Google’s Data Center Application (Paper Name omitted to maintain anonymity)

Data center

Abstract

As data centers provide more and more value to life, low efficiency and energy consumption of data center applications have become worldwide concerns. To address this concern,we comprehensively characterize micro-architectural implications of data center applications using newly released Google traces. We study the impact of thread switching, measure cache performance, and evaluate state-of-the-art optimization techniques to provide insights for optimizing data center ap-plications. We also investigate the similarity and diversity among traces and compare them with open-source widely-used applications to create a small representative subset that will facilitate future micro-architectural research. We will open-source the artifacts of this research

Publication
Peer-reviewed Poster. Paper in Submission to ASPLOS