簡易檢索 / 詳目顯示

研究生: 莊祐遠
You-Yuan Zhuang
論文名稱: SplitLSM: 在分區儲存固態硬碟上的LSM樹鍵值存儲的區域感知資料分配
SplitLSM: A Zone-Aware Data Allocation for LSM Tree-based Key-Value Store on Zoned Namespace SSDs
指導教授: 謝仁偉
Jen-Wei Hsieh
口試委員: 陳奕伸
Yi-shen Chen
陳雅淑
Ya-Shu Chen
張原豪
Yuan-Hao Chang
學位類別: 碩士
Master
系所名稱: 電資學院 - 資訊工程系
Department of Computer Science and Information Engineering
論文出版年: 2023
畢業學年度: 111
語文別: 英文
論文頁數: 33
中文關鍵詞: 分區儲存固態硬碟LSM樹鍵值存儲ZenFS寫入放大
外文關鍵詞: Zoned Namespace SSDs, LSM Tree-based Key-Value Store, ZenFS, Write amplification
相關次數: 點閱:272下載:0
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報

在傳統的固態硬碟(SSDs)中,某些應用場景的性能受到硬體設計限制的制約。因此,出現了各種新的SSD架構,尤其是在需要頻繁寫操作的場景中,分區命名空間固態硬碟(Zoned Namespace SSDs)展現出優越的寫入性能並被廣泛採用。此外,在大規模數據庫中,基於日誌結構合併樹(LSM Trees)常被用作後端存儲,因為它們的數據結構特性使其能夠更有效地處理連續寫入數據。這些因素使得LSM Trees成為大規模數據庫或數據中心等應用的熱門選擇。

在我們的研究中,我們探索了LSM Tree與分區命名空間固態硬碟的整合,旨在充分利用它們各自的優勢,以實現增強的性能。我們提出了SplitLSM,它從數據放置和合併兩個方面對在ZNS SSDs上廣泛使用的RocksDB(一個基於LSM Tree的廣泛使用的數據庫)進行了優化。我們的實驗結果表明,SplitLSM通過利用ZNS的區域架構和LSM Tree的連續寫入特性,提高了RocksDB在ZNS SSDs上的寫入性能。


In traditional SSDs, performance in certain application scenarios has been limited by hardware design constraints. As a result, various new SSD architectures have been developed, especially in scenarios that require frequent write operations, where Zoned Namespace SSDs have demonstrated outstanding write performance and are widely adopted. Additionally, in large-scale databases, Log-Structured Merge Trees (LSM Trees) are commonly used as backend storage due to their data structure characteristics, which allow them to handle continuous write data more effectively. These factors make LSM Trees a popular choice for applications such as large-scale databases or data centers.

In our research, we explore the integration of LSM Tree with ZNS SSDs with the goal of leveraging their respective strengths to achieve enhanced performance. We propose SplitLSM, which optimizes RocksDB (a widely used LSM Tree-based database) on ZNS SSDs from two aspects: data placement and compaction. Our experimental results demonstrate that SplitLSM improves the write performance of RocksDB on ZNS SSDs by harnessing the benefits of ZNS's zone architecture and LSM Tree's sequential write characteristics.

1 Introduction 2 Background 2.1 Zone namespace SSD 2.2 LSM-Tree based Key-value Store 2.3 Compaction 3 Motivation 3.1 ZenFS 3.2 Observation 3.3 Motivation 4 Method 4.1 Overview 4.2 Level splitting 4.2.1 Split strategy 4.2.2 Split the LSM tree 4.3 Zone-aware compaction 4.3.1 Temporary SSTable 5 Experiments 5.1 Experiment Setup 5.2 Experiment Results 5.2.1 Write performance 5.2.1 Compaction reduction 5.2.1 Compaction time 5.2.1 Compaction count 6 Conclusion

Matias Bjørling, Javier Gonzalez, and Philippe Bonnet. “LightNVM: The
Linux Open-Channel SSD Subsystem”. In: 15th USENIX Conference on
File and Storage Technologies (FAST 17). Santa Clara, CA: USENIX Association,
Feb. 2017, pp. 359–374. isbn: 978-1-931971-36-2.

Matias Bjørling et al. “ZNS: Avoiding the Block Interface Tax for Flashbased
SSDs”. In: 2021 USENIX Annual Technical Conference (USENIX
ATC 21). USENIX Association, July 2021, pp. 689–703.

Hao Chen et al. “SpanDB: A Fast, Cost-Effective LSM-tree Based KV
Store on Hybrid Storage”. In: 19th USENIX Conference on File and Storage
Technologies (FAST 21). USENIX Association, Feb. 2021, pp. 17–32.

Shuo-Han Chen, Yuhong Liang, and Ming-Chang Yang. “KVSTL: An
Application Support to LSM-Tree based Key-Value Store via Shingled
Translation Layer Data Management”. In: IEEE Transactions on Computers
(2021)

Yifan Dai et al. “From WiscKey to Bourbon: A Learned Index for Log-
Structured Merge Trees”. In: 14th USENIX Symposium on Operating Systems
Design and Implementation (OSDI 20). USENIX Association, Nov.
2020

Hee-Rock Lee et al. “Compaction-Aware Zone Allocation for LSM Based
Key-Value Store on ZNS SSDs”. In: Proceedings of the 14th ACM Workshop
on Hot Topics in Storage and File Systems. HotStorage ’22. Virtual
Event: Association for Computing Machinery, 2022

Yuhong Liang, Tsun-Yu Yang, and Ming-Chang Yang. “KVIMR: Key-
Value Store Aware Data Management Middleware for Interlaced Magnetic
Recording Based Hard Disk Drive”. In: 2021 USENIX Annual Technical
Conference (USENIX ATC 21). USENIX Association, July 2021

Russell Sears and Raghu Ramakrishnan. “BLSM: A General Purpose Log
Structured Merge Tree”. In: Proceedings of the 2012 ACM SIGMOD International
Conference on Management of Data. SIGMOD ’12. Scottsdale,
Arizona, USA: Association for Computing Machinery, 2012

Denghui Wu et al. “ZNSKV: Reducing Data Migration in LSMT-Based
KV Stores on ZNS SSDs”. In: 2022 IEEE 40th International Conference
on Computer Design (ICCD). 2022

Ting Yao et al. “GearDB: A GC-free Key-Value Store on HM-SMR Drives
with Gear Compaction”. In: 17th USENIX Conference on File and Storage
Technologies (FAST 19). Boston, MA: USENIX Association, Feb. 2019

Ting Yao et al. “MatrixKV: Reducing Write Stalls and Write Amplification
in LSM-tree Based KV Stores with Matrix Container in NVM”. In:
2020 USENIX Annual Technical Conference (USENIX ATC 20). USENIX
Association, July 2020

ZenFS. https : / / github . com / westerndigitalcorporation / zenfs /
tree/master. 2023

zonedatorage. https://zonedstorage.io/.2023.

無法下載圖示 全文公開日期 2025/08/29 (校內網路)
全文公開日期 2025/08/29 (校外網路)
全文公開日期 2025/08/29 (國家圖書館:臺灣博碩士論文系統)
QR CODE