AegonKV: A High Bandwidth, Low Tail Latency, and Low Storage Cost KV-Separated LSM Store
summary
AegonKV moves garbage-collection for KV-separated LSM storage onto a Samsung first-generation SmartSSD-based GC offloading, implementing a bitmap valid map and a host-side GC unit (GCU) plus a lightweight GC Computation Unit on the device to run asynchronous GC without stealing host CPU or host I/O bandwidth. Experiments with YCSB and production Facebook social-graph and Twitter cluster traces show 1.28–3.3× throughput, 37–66% lower tail latency, and 15–85% less space overhead versus prior KV-separated systems.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:22GC in KV separation challenges
- 4:55Smart SSD near-data processing
- 6:00AegonKV design
- 9:58GC on Smart SSD hardware
- 10:19Evaluation and benchmarks
- 13:55Parameter sensitivity
- 14:36Summary and conclusion
related talks
18mSplitKV: Splitting IO Paths for Different Sized Key-Value Items with Advanced
12mSpanDB: A Fast, Cost-Effective LSM-tree Based KV Store on Hybrid Storage
21mMatrixKV: Reducing Write Stalls and Write Amplification in LSM-tree Based KV Stores
18mPreview: KV-Stores and Databases
15mMitigating Resource Usage Dependency in Sorting-based KV Stores on Hybrid...
14mLoLKV: The Logless, Linearizable, RDMA-based Key-Value Storage System
13mDifferentiated Key-Value Storage Management for Balanced I/O Performance
28mGearDB: A GC-free Key-Value Store on HM-SMR Drives with Gear Compaction