Linux System Telemetry Dataset for Memory Pressure and Resource Contention Anomaly Detection (Version 1.1)

This dataset contains multivariate system-level telemetry collected from a Linux server over a continuous 20-day monitoring period. Data were recorded at fixed 60-second intervals between 27 July 2025 (17:41:58) and 16 August 2025 (22:37:57). The dataset consists of 29,095 time-stamped observations. Each observation includes 13 numerical system metrics describing CPU utilisation, memory usage, buffer and cache behaviour, and swap activity. The dataset contains both baseline system behaviour and controlled workload-induced anomaly periods generated through bounded memory pressure and high-concurrency request bursts. This enables reproducible research in anomaly detection and resource contention analysis. The dataset is suitable for: Anomaly detection Predictive system monitoring Resource contention analysis Time-series modelling Explainable AI for system diagnostics Version 1.1: Repository structure updated. Dataset files are provided individually rather than as a compressed archive. The README documentation has been revised for clarity and accuracy. No changes were made to the underlying dataset content.本数据集包含在连续20天监控周期内从Linux服务器采集的多变量系统级遥测数据 (multivariate system-level telemetry)。数据采集于2025年7月27日 (17:41:58) 至2025年8月16日 (22:37:57) 期间，以固定60秒间隔进行记录。 本数据集共包含29095条带时间戳的观测记录。每条记录包含13项数值型系统指标，涵盖CPU利用率 (CPU utilisation)、内存使用 (memory usage)、缓冲区与缓存 (buffer and cache) 行为以及交换空间活动 (swap activity)。 本数据集同时包含基线系统行为，以及通过有限内存压力 (bounded memory pressure) 和高并发请求突发 (high-concurrency request bursts) 生成的受控工作负载 (workload) 诱发的异常时段，可支撑异常检测与资源争用分析领域的可复现研究。 本数据集适用于： - 异常检测 - 预测性系统监控 - 资源争用分析 - 时间序列建模 (time-series modelling) - 面向系统诊断的可解释AI (Explainable AI) 版本1.1：更新了仓库结构 (repository structure)，数据集文件以单文件形式单独提供，而非压缩归档包 (compressed archive)。README文档已修订以提升清晰度与准确性，数据集核心内容未做任何修改。