An Analysis of HPC Benchmarks in Virtual Machine Environments

Abstract

Abstract. Virtualization technology has been gaining acceptance in the scientific community due to its overall flexibility in running HPC applications. It has been reported that a specific class of applications is better suited to a particular type of virtualization scheme or implementation. For example, Xen has been shown to perform with little overhead for compute-bound applications. Such a study, although useful, does not allow us to generalize conclusions beyond the performance analysis of that application which is explicitly executed. An explanation of why the generalization described above is difficult, may be due to the versatility in applications, which leads to different overheads in virtual environments. For example, two similar applications may spend disproportionate amount of time in their respective library code when run in virtual environments. In this paper, we aim to study such potential causes by investigating the behavior and identifying patterns of various overheads for HPC benchmark applications. Based on the investigation of the overhead profiles for different benchmarks, we aim to address questions such as: Are the overhead profiles for a particular type of benchmarks (such as compute-bound) similar or are there grounds to conclude otherwise?

Evaluate the Performance and Scalability of Image Deployment in Virtual Data Center

“…Many researchers put their focus on the performance overheads of several pivotal system components in virtualization environments, e.g., CPU, memory, disk and network etc [5,6]. Some researchers focus the performance overheads in specific scenarios, such as server consolidation [7,8] and HPC environment [9]. However, to the best of our knowledge, none of them have quantified the deployment efficiency for typical server applications and HPC applications under different deployment strategies.…”

DVFS Governor for HPC: Higher, Faster, Greener

“…But even with low-overhead virtualization techniques for HPC systems [16], [26], [28] consolidation is of no use on fully loaded systems.…”

Scheduling Medical Application Workloads on Virtualized Computing Systems

“…Noting that virtualization adds computational overhead, as we will describe in Chapter III, others have studied the performance of virtualized applications [5], [19][20][21][22][23][24]; some of them specifically studied scientific applications [20][21][22][23][24][25]. For example, in [20], the authors evaluate the performance impact of Xen on different parallel application benchmarks.…”

“…In [23], they use some popular parallel benchmarks to evaluate the overhead of Xen. They use more low-level profiling details such as the number of cache misses.…”

“…A drawback with virtualization is that it adds computational overhead [5,[20][21][22][23][24][25]. For scientific applications in particular, virtualization has been shown to result in a significant performance penalty.…”