Overview
Explore Facebook's innovative approach to scaling memcached in this USENIX conference talk from NSDI '13. Discover how the social media giant leverages memcached as a building block to construct and scale a distributed key-value store supporting billions of requests per second for over a billion users worldwide. Learn about infrastructure and design requirements, demand fill strategies, handling stale sets and thundering herds, and techniques for scaling across multiple data centers. Gain insights into Facebook's solutions for database invalidation, reducing packet density, and implementing remote markers. Benefit from the lessons learned by Facebook's engineering team as they tackled the challenges of scaling memcache to unprecedented levels.
Syllabus
Introduction
Infrastructure Requirements
Design Requirements
Memcache
Background
Why Memcache
Memcache Servers
Demand Fill
Delete
Stale Sets
Thundering Herds
Scaling
Frontend Cluster
Database Invalidation
Reducing Packet Density
Larger Scale
Distributed Data Centers
Single Master
NonMaster
Remote Markers
Recap
Lessons Learned
Questions
Clarification
Taught by
USENIX