Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Nezha: A Key-Value Separated Distributed Store with Optimized Raft Integration

Yangyang Wang, Yucong Dong, Ziqian Cheng, Zichen Xu

TL;DR

Nezha is presented, a prototype distributed storage system that innovatively integrates key-value separation with Raft to provide scalable throughput in a strong consistency guarantee and incorporates leveled garbage collection, significantly improving read and write performance while preserving Raft's safety properties.

Abstract

Distributed key-value stores are widely adopted to support elastic big data applications, leveraging purpose-built consensus algorithms like Raft to ensure data consistency. However, through systematic analysis, we reveal a critical performance issue in such consistent stores, i.e., overlapping persistence operations between consensus protocols and underlying storage engines result in significant I/O overhead. To address this issue, we present Nezha, a prototype distributed storage system that innovatively integrates key-value separation with Raft to provide scalable throughput in a strong consistency guarantee. Nezha redesigns the persistence strategy at the operation level and incorporates leveled garbage collection, significantly improving read and write performance while preserving Raft's safety properties. Experimental results demonstrate that, on average, Nezha achieves throughput improvements of 460.2%, 12.5%, and 72.6% for put, get, and scan operations, respectively.

Nezha: A Key-Value Separated Distributed Store with Optimized Raft Integration

TL;DR

Nezha is presented, a prototype distributed storage system that innovatively integrates key-value separation with Raft to provide scalable throughput in a strong consistency guarantee and incorporates leveled garbage collection, significantly improving read and write performance while preserving Raft's safety properties.

Abstract

Distributed key-value stores are widely adopted to support elastic big data applications, leveraging purpose-built consensus algorithms like Raft to ensure data consistency. However, through systematic analysis, we reveal a critical performance issue in such consistent stores, i.e., overlapping persistence operations between consensus protocols and underlying storage engines result in significant I/O overhead. To address this issue, we present Nezha, a prototype distributed storage system that innovatively integrates key-value separation with Raft to provide scalable throughput in a strong consistency guarantee. Nezha redesigns the persistence strategy at the operation level and incorporates leveled garbage collection, significantly improving read and write performance while preserving Raft's safety properties. Experimental results demonstrate that, on average, Nezha achieves throughput improvements of 460.2%, 12.5%, and 72.6% for put, get, and scan operations, respectively.
Paper Structure (32 sections, 11 figures, 2 tables, 3 algorithms)

This paper contains 32 sections, 11 figures, 2 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. BACKGROUND AND MOTIVATION
  3. Consensus Algorithms and Data Persistence
  4. Raft protocol
  5. Key-Value Separation
  6. Motivation
  7. NEZHA
  8. Overview
  9. Optimized Raft Log Storage Mechanism
  10. Adaptive Storage Management Mechanism
  11. Three-Phase Request Processing Mechanism
  12. Put Request Processing
  13. Get Request Processing
  14. Scan Request Processing
  15. Fault Tolerance and Safety Guarantees
  16. ...and 17 more sections

Figures (11)

  • Figure 1: An example of the request processing of Raft.
  • Figure 2: Processing flow of write operations in a Raft-based distributed key-value store with LSM-tree storage engine.
  • Figure 3: The architecture of Nezha.
  • Figure 4: Performance comparisons for put requests under different value sizes.
  • Figure 5: Performance comparisons for point queries under different value sizes.
  • ...and 6 more figures