SplitKV: Splitting IO Paths for Different Sized Key-Value Items with Advanced
Talkat 12th USENIX Workshop on Hot Topics in Storage and File Systems (HotStorage '20) · 2020-07-13..07-14·
summary
SplitKV is a prototype key-value store that writes small KV to PM then migrates to SSD and directly writes large KV to SSD, using Intel DC persistent memory plus NVMe SSDs and a global in-memory B+ tree index. It picks a 4 KB cutoff (≤4 KB → PM path) and reclaims PM by migrating cold items based on simple access-count hotness. Evaluation versus RocksDB, KVell and NoveLSM shows up to 3.5× throughput generally and up to 7.9× under Zipf-like mixes, with lower latencies for mixed small/large workloads.
AI-written summary · about
chapters (5)
related talks
13mDifferentiated Key-Value Storage Management for Balanced I/O Performance
19mBuilding KVStore for ML Workloads
21mMatrixKV: Reducing Write Stalls and Write Amplification in LSM-tree Based KV Stores
15mMitigating Resource Usage Dependency in Sorting-based KV Stores on Hybrid...
12mSpanDB: A Fast, Cost-Effective LSM-tree Based KV Store on Hybrid Storage
15mKVIMR: Key-Value Store Aware Data Management Middleware for Interlaced Magnetic...
32mKeyvi the Key Value Index Optimized for Size and Speed #bbuzz
16mAegonKV: A High Bandwidth, Low Tail Latency, and Low Storage Cost KV-Separated LSM Store
