DP-HPC: Bringing Differential Privacy to HPC Systems Log Sharing and Analysis
Ana Luisa V. Solorzano · OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information) · 2025
Monitoring HPC systems yields valuable insights into user behavior, aiding resource management, collaborative research, and software design.However, privacy concerns restrict continuous log sharing.Traditional anonymization methods fall short as user behavior remains identifiable.To address this, we propose a robust toolset for preserving HPC system logs' privacy using Differential Privacy (DP), ensuring released values reveal no individual information.Our toolset applies configurable random noise to data, accompanied by metrics and visualizations for making informed privacy decisions.Our tool is effective in tuning the privacy of system logs collected over time.Moreover, models trained on privacy-preserved logs maintain accuracy compared to real data.•