Scaling Memcache at Facebook — How Facebook’s Caching System Handles Billions of Requests

Facebook’s infrastructure serves billions of requests every second — and behind that speed is Memcache, a distributed, high-performance caching system built on top of the open-source Memcached.

In this video, we break down Facebook’s technical paper “Scaling Memcache at Facebook”, explaining how engineers at Facebook redesigned and scaled Memcached into a massive, multi-region distributed system.

We explore how Facebook handles caching at three levels:
Within a single server cluster — optimizing performance with fine-grained locking and an adaptive slab allocator.
Across multiple clusters in a region — using leases to prevent stale sets and reduce “thundering herds.”
Across regions globally — managing consistency and replication with the mcsqueal daemon and remote markers.

You’ll also learn how Facebook’s Gutter system isolates backend services during partial outages, keeping the site running even when infrastructure components fail.

From billions of reads per second to microsecond-level latency, this is how Facebook built one of the most resilient and scalable caching systems ever engineered.
#FacebookEngineering #DistributedSystems #Memcache #Memcached #Caching #Scalability #DataEngineering #SystemDesign #TechExplained #ComputerScience #Networking #BackendEngineering #Infrastructure #FacebookTech #WebPerformance #techeducation

Видео Scaling Memcache at Facebook — How Facebook’s Caching System Handles Billions of Requests канала Nabadeep Das