AC-Key: Adaptive Caching for LSM-based Key-Value Stores
summary
AC-Key implements an adaptive cache inside RocksDB that manages key-value cache, key-pointer cache, and block cache, resizing them via hierarchical ghost caches and a novel caching efficiency factor that weighs per-entry cost and benefit. It is built by modifying RocksDB and evaluated against default RocksDB configs and an exhaustive offline fixed-size tuner. Results show higher throughput across mixed workloads and it even beats the best offline fixed-size configuration during abrupt phase changes.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 0:57Internal Operations
- 13:33Hierarchical Adaptive Caching
- 16:57Challenges
- 17:46Implementation
- 20:19Phase Change Test
- 23:18Summary
related talks
29mAccelerating Performance: Mastering RocksDB for High Speed Data Processing
11mEvolution of Development Priorities in Key-value Stores Serving Large-scale Applications
15mADOC: Automatically Harmonizing Dataflow Between Components in Log-Structured Key-Value
24mARKDB: The Key-Value Engine for Alibaba Cloud Storage Services
15mImproving Performance of Flash Based Key-Value Stores Using Storage Class Memory...
54mEVCache: Lowering Costs for a Low-Latency Cache with RocksDB
19mA zone-aware cache system for distributed databases
20mSymbiosis: The Art of Application and Kernel Cache Cooperation