Evaluating Interconnect and Virtualization Performance forHigh Performance Computing

  • Authors:
  • Lavanya Ramakrishnan;R. Shane Canon;Krishna Muriki;Iwona Sakrejda;Nicholas J. Wright

  • Affiliations:
  • Lawrence Berkeley National Lab, Berkeley, CA;Lawrence Berkeley National Lab, Berkeley, CA;Lawrence Berkeley National Lab, Berkeley, CA;Lawrence Berkeley National Lab, Berkeley, CA;Lawrence Berkeley National Lab, Berkeley, CA

  • Venue:
  • ACM SIGMETRICS Performance Evaluation Review
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Scientists are increasingly considering cloud computing platforms to satisfy their computational needs. Previous work has shown that virtualized cloud environments can have significant performance impact. However there is still a limited understanding of the nature of overheads and the type of applications that might do well in these environments. In this paper we detail benchmarking results that characterize the virtualization overhead and its impact on performance. We also examine the performance of various interconnect technologies with a view to understanding the performance impacts of various choices. Our results show that virtualization can have a significant impact upon performance, with at least a 60% performance penalty. We also show that less capable interconnect technologies can have a significant impact upon performance of typical HPC applications. We also evaluate the performance of the Amazon Cluster compute instance and show that it performs approximately equivalently to a 10G Ethernet cluster at low core counts.