Power-aware dynamic placement of HPC applications

TLDR

High‑performance computing platforms have traditionally ignored power consumption, but rising energy costs have made power management a critical concern for compute‑intensive server clusters. The study investigates applying power‑management techniques to high‑performance applications on modern, power‑efficient, virtualized servers. The authors evaluate dynamic consolidation and low‑power state utilization, conduct a comprehensive experimental study to assess application isolation, and develop a framework for power‑aware placement of HPC workloads on virtualized servers. They find that virtualization overhead and application isolation are major bottlenecks, that HPC power consumption is application‑dependent, non‑linear, and highly variable, and that working‑set size is a critical factor for placement decisions.

Abstract

High Performance Computing applications and platforms have been typically designed without regard to power consumption. With increased awareness of energy cost, power management is now an issue even for compute-intensive server clusters. In this work, we investigate the use of power management techniques for high performance applications on modern power-efficient servers with virtualization support. We consider power management techniques such as dynamic consolidation and usage of dynamic power range enabled by low power states on servers. We identify application performance isolation and virtualization overhead with multiple virtual machines as the key bottlenecks for server consolidation. We perform a comprehensive experimental study to identify the scenarios where applications are isolated from each other. We also establish that the power consumed by HPC applications may be application dependent, non-linear and have a large dynamic range. We show that for HPC applications, working set size is a key parameter to take care of while placing applications on virtualized servers. We use the insights obtained from our experimental study to present a framework and methodology for power-aware application placement for HPC applications.