Per-cgroup Swap Device Control
summary
Proposal implements per-cgroup swap tiers so cgroups pick from named groups of swap devices mapped to priority ranges (example ranges: network -1–40, HDD 41–100, SSD 101–SHORT_MAX), and adds two selection primitives: a parent-inheritance op (inherit / ignore-parent-start / ignore-parent-zero) and per-tier ops (plus to select, minus to exclude). Patches exist as RFC/prototype and raise concrete implementation questions: percpu and fast‑path changes, NUMA autobind exclusivity, a possible swap.tier.max, parent–child recharging/migration semantics, and whether unmapped pages must be migrated. inherit/override rules and tier mask calculation remain unresolved.
AI-written summary · about
chapters (6)
- 0:00Introduction
- 0:21Background and motivation
- 1:44Swap tier concept explained
- 3:11Per-cgroup tier control rules
- 5:18Discussion and questions
- 14:19Wrap-up and break
related talks
14mScaleSwap: A Scalable OS Swap System for All-Flash Swap Arrays
26mCXL Memory Tiering for heterogenous computing
38mUnderstanding and Working with the Cgroups Interface
31mHigh Density VMs in OpenShift: The Journey to Native K8s Swap
45mLinux Kernel Control Group v2
27msystemd cgroup delegation and control processes
57mAn introduction to control groups (cgroups) version 2
35mFuture-Proof Your Applications: Avoid Breakdowns With Cgroup Chan...