DiFache: Efficient and Scalable Caching on Disaggregated Memory using Decentralized Coherence

TL;DR

DiFache introduces a decentralized, scalable caching framework for disaggregated memory systems that reduces coherence overhead and improves performance by aligning cache consistency with memory nodes.

Contribution

It proposes a novel decentralized coherence mechanism and adaptive caching scheme tailored for disaggregated memory architectures, enhancing scalability and efficiency.

Findings

Up to 10.83× performance improvement over existing methods

Average 5.53× performance gain across evaluations

Peak throughput increases of 7.94× and 2.19× for real-world applications

Abstract

The disaggregated memory (DM) architecture offers high resource elasticity at the cost of data access performance. While caching frequently accessed data in compute nodes (CNs) reduces access overhead, it requires costly centralized maintenance of cache coherence across CNs. This paper presents DiFache, an efficient, scalable, and coherent CN-side caching framework for DM applications. Observing that DM applications already serialize conflicting remote data access internally rather than relying on the cache layer, DiFache introduces decentralized coherence that aligns its consistency model with memory nodes instead of CPU caches, thereby eliminating the need for centralized management. DiFache features a decentralized invalidation mechanism to independently invalidate caches on remote CNs and a fine-grained adaptive scheme to cache objects with varying read-write ratios. Evaluations using 54 real-world traces from Twitter show that DiFache outperforms existing approaches by up to 10.83$\times$ (5.53$\times$ on average). By integrating DiFache, the peak throughput of two real-world DM applications increases by 7.94$\times$ and 2.19$\times$, respectively.