用于调查计算集群异常的数据集

该数据集由托马斯·杰斐逊国家加速器实验室创建，旨在通过收集332个计算节点的数据来研究计算集群的正常与异常行为。数据集涵盖了从2023年5月19日至23日的数据，其中5月19日至22日代表正常行为，而5月23日记录了一次显著的异常事件。数据集包含多种硬件配置，详细记录了CPU、磁盘、内存和Slurm的性能指标，总数据量超过180GB。此数据集适用于开发无监督机器学习算法，以检测计算集群中的异常事件。

This dataset was created by the Thomas Jefferson National Accelerator Facility to study the normal and anomalous behaviors of computing clusters by collecting data from 332 compute nodes. It covers data from May 19 to 23, 2023, where May 19 to 22 represent normal operational behaviors, while a notable anomalous event was recorded on May 23. The dataset includes various hardware configurations, and comprehensively records performance metrics of CPU, disk, memory, and Slurm, with a total data volume exceeding 180 GB. This dataset is suitable for developing unsupervised machine learning algorithms to detect anomalous events in computing clusters.