Linux Memory Management at Scale - Under the Hood

Explore Linux memory management at scale in this 53-minute conference talk by Chris Down from Facebook at SREcon19 Asia/Pacific. Dive deep into fundamental concepts crucial for site reliability engineers, including swap, cgroup v2, and memory protection. Discover how to build reliable memory-heavy systems, debunk common misconceptions, and learn about cutting-edge kernel concepts like PSI. Gain insights into scaling concerns, priority inversions, different types of memory, and the OOM Killer. Understand reclaim processes, physical memory limitations, and pre-OOM detection. Test your knowledge with a pop quiz and explore memory limits, roles, and protection in production environments. Enhance your ability to create stable and performant systems at scale through this comprehensive exploration of Linux memory management.

Intro
Overview
Talk within a talk
Scaling concerns
Why Cgroups
Priority inversions
Ext4 Priority inversions
Different types of memory
Swap
Ome Killer
How does reclaim work
Running out of physical memory
Pre ohm detection
Pop Quiz
Memory Limits
EveryText
Base OS Layer
Roles
Protection
Production Success