Segcache: a memory-efficient and scalable in-memory key-value cache for small objects
summary
Segment-structured cache that groups small objects by similar creation/expiration into fixed-size segments, lifts most per-object metadata into shared segment headers and a shared hash-table slot, and expires/evicts at segment granularity with tiny critical sections. Implemented on Pelican and evaluated on Twitter production traces, it uses 22–60% less memory than state of the art, delivers up to 40% better single-thread throughput and shows ≈8× speedup with 24 threads versus Memcached while keeping near-zero fragmentation.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 1:21Time to Need
- 2:39Existing Solutions
- 3:41Motivation Summary
- 4:45Overview
- 5:12Components
- 6:34Design principles
- 8:38Macro management
- 9:30Evaluation
- 10:01Memory efficiency
- 10:42Stability
- 11:12Summary
related talks
49mCache à la carte: a framework for in-memory caching
25mMemC3: Compact and Concurrent MemCache with Dumber Caching and Smarter Hashing
23mScaling Memcache
17mScalaCache: Scalable User-Space Page Cache Management with Software-Hardware...
24mWrite-Optimized Dynamic Hashing for Persistent Memory
27mEfficient shared memory data structures
43mCaching half a billion user transactions
42mCaching for Cash