Optimizing Space Amplification in RocksDB
Summary: Introduces practical techniques to minimize space amplification in RocksDB—trading CPU and read/write amplification via compaction/tuning—to prioritize storage efficiency while preserving OLTP latency targets. Empirical (TPC-C, LinkBench, production) results show RocksDB uses <50% of InnoDB storage and often matches or exceeds its performance, demonstrating the first large-scale competitive LSM deployment for OLTP. (summarized by gpt-5-mini on Feb 09 2026)
Incoming Non-self Citations Over Time
Authors
- 1. Siying Dong (Meta)
- 2. Mark Callaghan (Meta)
- 3. Leonidas Galanis (Meta)
- 4. Dhruba Borthakur (Meta)
- 5. Tony Savor (Meta)
- 6. Michael Stumm (University of Toronto)
BibTeX Citation
@inproceedings{dong_cidr17,
address = {Amsterdam, Netherlands},
series = {{CIDR} '17},
title = {{Optimizing Space Amplification in RocksDB}},
booktitle = {Proceedings of the {Conference} on {Innovative} {Data} {Systems} {Research}},
author = {Dong, Siying and Callaghan, Mark and Galanis, Leonidas and Borthakur, Dhruba and Savor, Tony and Stumm, Michael},
year = {2017}
}Incoming Citations (Sorted by Pagerank)
Showing 1 of 51 citing papers.
| Rank | Citing Paper | Year | Venue | Pagerank |
|---|---|---|---|---|
| 11,372 | Optimal Uncoordinated Unique IDs | 2023 | PODS | 5.093636e-05 |
Outgoing Citations (Sorted by Pagerank)
Showing 5 of 5 cited papers.
Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.
| Rank | Cited Paper | Year | Venue | Pagerank |
|---|---|---|---|---|
| 232 | LinkBench: a Database Benchmark Based on the Facebook Social Graph | 2013 | SIGMOD | 0.00023830978 |
| 275 | bLSM: A General Purpose Log Structured Merge Tree | 2012 | SIGMOD | 0.00022630189 |
| 1,389 | Realtime Data Processing at Facebook | 2016 | SIGMOD | 0.00010943767 |
| 2,805 | LogBase: A Scalable Log-structured Database System in the Cloud | 2012 | VLDB | 8.1044134e-05 |
| 2,906 | Optimizing Optimistic Concurrency Control for Tree-Structured, Log-Structured Databases | 2015 | SIGMOD | 7.9741374e-05 |
Previous
Page 1 / 1
Next
Semantically Similar Papers
| # | Overall Rank | Paper | Year | Venue |
|---|---|---|---|---|
| 1 | 9,465 | Are Joins over LSM-trees Ready? Take RocksDB as an Example | 2025 | VLDB |
| 2 | 5,080 | Hybrid Storage Management for Database Systems | 2013 | VLDB |
| 3 | 1,580 | Compaction management in distributed key-value datastores | 2015 | VLDB |
| 4 | 10,465 | Improving Range Scan Performance in LSM-trees with Group Caching | 2026 | SIGMOD |
| 5 | 3,773 | SplinterDB and Maplets: Improving the Tradeoffs in Key-Value Store Compaction Policy | 2023 | SIGMOD |
| 6 | 4,894 | Optimizing Databases by Learning Hidden Parameters of Solid State Drives | 2020 | VLDB |
| 7 | 857 | SlimDB: A Space-Efficient Key-Value Storage Engine For Semi-Sorted Data | 2017 | VLDB |
| 8 | 9,950 | CloudJump: Optimizing Cloud Databases for Cloud Storages | 2022 | VLDB |
| 9 | 4,200 | Disaggregating RocksDB: A Production Experience | 2023 | SIGMOD |
| 10 | 1,213 | MyRocks: LSM-Tree Database Storage Engine Serving Facebook's Social Graph | 2020 | VLDB |