DAMON Current Status and Future Plans
summary
DAMON provides page-granularity sampling (defaults: one page per region, NR=10, sample=5ms, aggregation=100ms) and DAMON-based operation schemes (DAMOS) that apply actions like page-out or THP with a quota (example: limit e.g. 100 MB/s). A service reported about a 16% reduction in memory usage with modest overhead. Recent merges add TR-region export (6.2), filtering by backing/cgroup/address (6.3 then 6.6), and moving-average / dapply timing (6.7-rc1). Autotuning RFC is posted and shows online tuning reaching offline-tuned PSI goals; future work covers AKMA stealing, working-set metrics, and tiered-memory placement.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:27Why DAMON exists
- 4:20DAMON basics
- 13:30DAMON updates since last year
- 21:04Future plan: AI-driven tuning
- 26:47Future plan: Access continuity
- 42:45Future plan: Auto-tiered memory
- 46:03Q&A and conclusion
related talks
45mCurrent Status and Future Plans of DAMON
52mDAMON, DAMOS, and DAMO: Kernel Subsystems and User-Space Tools for Data Access...
42mDAMON Recipes: Ways to Save Memory Using a Linux Kernel Subsystem in the Real World- 42mDAMON Recipes: Ways to Save Memory Using a Linux Kernel Subsystem in the Real World
48mData Access Monitoring Operator (DAMO): User-Space Tool/Python Library for Access-Aware Profiling and Optimization of Your Linux Systems
47mDAMO[N,S]?: Implementing Self-Driven Data Access-Aware Efficient Linux System
31mHMSDK: Enhancing Software Support for CXL Memory
43mLive in a world with multiple memory types