A Review of Supercomputer Performance Monitoring Systems

Abstract

High Performance Computing is now one of the emerging fields in computer science and its applications. Top HPC facilities, supercomputers, offer great opportunities in modeling diverse processes thus allowing to create more and greater products without full-scale experiments. Current supercomputers and applications for them are very complex and thus are hard to use efficiently. Performance monitoring systems are the tools that help to understand the efficiency of supercomputing applications and overall supercomputer functioning. These systems collect data on what happens on a supercomputer (performance data, performance metrics) and present them in a way allowing to make conclusions about performance issues in programs running on the supercomputer. In this paper we give an overview of existing performance monitoring systems designed for or used on supercomputers. We give a comparison of performance monitoring systems found in literature, describe problems emerging in monitoring large scale HPC systems, and outline our vision on future direction of HPC monitoring systems development.

In-Network Monitoring Strategies for HPC Cloud

“…Monitoring RDMA networks requires specialized tools and techniques designed to capture and analyze RDMA-specific traffic and performance metrics. These tools provide insights into key parameters such as bandwidth utilization, latency, and congestion [19]. In the following, some of the known tools are discussed:…”

Analysing Supercomputer Nodes Behaviour with the Latent Representation of Deep Learning Models

“…The scale of the problem motivates the development of automated procedures for anomaly detection and faulty node identification in current supercomputers and this need will become even more pressing for future Exascale systems [6]. The fact that most of today's HPC computing systems are endowed with monitoring infrastructures [7] that gather data from software (SW) and hardware (HW) components can be of great help toward the development of data-driven automated approaches. Historically, system management was performed through hand-crafted scripts and direct intervention of system administrators; most of the data is stored in log files, and anomalies are investigated a posteriori to find the source of reported problems (e.g., when many users recognize the failure and report it to administrators).…”