A Lock-free RDMA-friendly Index in CPU-parsimonious Environments

In CPU-parsimonious environments, such as disaggregated memory systems, the limited CPU power on the memory side constrains the ability to perform more operations. Thus, reducing CPU usage and enhancing concurrency performance are critical for indexing key-value storage in these scenarios. Current hash indexes support one-sided RDMA access and lock-free concurrency control with good performance but lack range query support. In contrast, existing tree indexes support range query but only implement expensive locks for concurrency control. Therefore, designing a tree index that supports lock-free concurrent access and one-sided RDMA is a significant challenge. To address these issues, this paper proposes a lock-free tree index based on the van Emde Boas (vEB) tree, called vBoost. vBoost inherits the vEB tree’s characteristics of index nodes without splitting or merging, and simplifies concurrency control by managing changes at a single node. It redesigns tree nodes with an 8-byte compact data structure, allowing each key-value pair to support RDMA_CAS atomic operations for lock-free concurrency. Additionally, vBoost leverages the RDMA Doorbell technique to reduce RTTs, enhancing range query and write performance. Evaluation results show that, vBoost achieves up to 3.85× higher throughput and better scalability under YCSB workloads compared to state-of-the-art tree indexes.

Authors

Institutions

Publication Details

Journal
ACM Transactions on Architecture and Code Optimization
Published
2025-03-07
DOI
https://doi.org/10.1145/3722112
Primary Topic
Distributed systems and fault tolerance
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

A Lock-free RDMA-friendly Index in CPU-parsimonious Environments

Yun Xu, Weiguang Wang, Yonghui Xu, Yuting Li et al.
ACM Transactions on Architecture and Code Optimization
Distributed systems and fault tolerance
article

A Lock-free RDMA-friendly Index in CPU-parsimonious Environments

Yun Xu, Weiguang Wang, Yonghui Xu, Yuting Li, Pengcheng Wang
article en

Abstract

In CPU-parsimonious environments, such as disaggregated memory systems, the limited CPU power on the memory side constrains the ability to perform more operations. Thus, reducing CPU usage and enhancing concurrency performance are critical for indexing key-value storage in these scenarios. Current hash indexes support one-sided RDMA access and lock-free concurrency control with good performance but lack range query support. In contrast, existing tree indexes support range query but only implement expensive locks for concurrency control. Therefore, designing a tree index that supports lock-free concurrent access and one-sided RDMA is a significant challenge. To address these issues, this paper proposes a lock-free tree index based on the van Emde Boas (vEB) tree, called vBoost. vBoost inherits the vEB tree’s characteristics of index nodes without splitting or merging, and simplifies concurrency control by managing changes at a single node. It redesigns tree nodes with an 8-byte compact data structure, allowing each key-value pair to support RDMA_CAS atomic operations for lock-free concurrency. Additionally, vBoost leverages the RDMA Doorbell technique to reduce RTTs, enhancing range query and write performance. Evaluation results show that, vBoost achieves up to 3.85× higher throughput and better scalability under YCSB workloads compared to state-of-the-art tree indexes.

ACM Transactions on Architecture and Code OptimizationVol. 22(2)
University of Science and Technology of China (CN), Huawei Technologies (China) (CN), Huawei Technologies (United Kingdom) (GB)
Openalex Percentile: Top 96%
Distributed systems and fault tolerance
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

A Lock-free RDMA-friendly Index in CPU-parsimonious Environments — Yun Xu, Weiguang Wang, et al. · ACM Transactions on Architecture and Code Optimization (2025) | TGRS Research Map | TGRS