What is it about?

Data centers are embracing the disaggregated memory (DM) architecture to improve resource efficiency and save costs. B+ trees are usually chosen to store and retrieve data on DM, but are they suitable for DM? This paper proposes the radix tree as a better choice and optimizes it to fit DM.

Featured Image

Why is it important?

Today's cloud data centers have long suffered from low memory utilization (< 60%). Disaggregated memory (DM) is an increasingly prevalent architecture proposed to address such an issue. However, as DM decouples the computing and memory resources, traditional indexing solutions do not work well. Our findings show that the performance of existing B+ trees on DM is up to 10.8 times lower than the theoretical upper bound.

Perspectives

This data structure is specifically designed for disaggregated memory. It also contains some useful and general designs, such as a read delegation and write combining design, a pure lock-free ART design. I hope you find this paper interesting.

Xuchuan Luo
Fudan University

This paper innovatively proposes to use radix trees as range index data structures on disaggregated memory (DM). This work can be viewed as a guideline on how to design efficient range indexes on DM.

Jiacheng Shen
The Chinese University of Hong Kong

Read the Original

This page is a summary of: A Memory-Disaggregated Radix Tree, ACM Transactions on Storage, May 2024, ACM (Association for Computing Machinery),
DOI: 10.1145/3664289.
You can read the full text:

Read

Resources

Contributors

The following have contributed to this page