Spatiotemporal Modeling of Node Temperatures in Supercomputers

• 成果类型: Article
• 署名作者: Storlie, Curtis B.; Reich, Brian J.; Rust, William N.; Ticknor, Lawrence O.; Bonnie, Amanda M.; Montoya, Andrew J.; Michalak, Sarah E.
• 署名单位: United States Department of Energy (DOE); Los Alamos National Laboratory; North Carolina State University; United States Department of Energy (DOE); Los Alamos National Laboratory
• 刊物名称: JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
• ISSN/ISSBN: 0162-1459
• DOI: 10.1080/01621459.2016.1195271
• 发表日期: 2017
• 页码: 92-108
• 关键词: univariate

摘要

Los Alamos National Laboratory is home to many large supercomputing clusters. These clusters require an enormous amount of power (approximate to 500-2000 kW each), and most of this energy is converted into heat. Thus, cooling the components of the supercomputer becomes a critical and expensive endeavor. Recently, a project was initiated to investigate the effect that changes to the cooling system in a machine room had on three large machines that were housed there. Coupled with this goal was the aim to develop a general good-practice for characterizing the effect of cooling changes and monitoring machine node temperatures in this and other machine rooms. This article focuses on the statistical approach used to quantify the effect that several cooling changes to the room had on the temperatures of the individual nodes of the computers. The largest cluster in the room has 1600 nodes that run a variety of jobs during general use. Since extremes temperatures are important, a Normal distribution plus generalized Pareto distribution for the upper tail is used to model the marginal distribution, along with a Gaussian process copula to account for spatio-temporal dependence. A Gaussian Markov random field (GMRF) model is used to model the spatial effects on the node temperatures as the cooling changes take place. This model is then used to assess the condition of the node temperatures after each change to the room. The analysis approach was used to uncover the cause of a problematic episode of overheating nodes on one of the supercomputing clusters. This same approach can easily be applied to monitor and investigate cooling systems at other data centers, as well. Supplementary materials for this article are available online.