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

2025-05-23

AI Summary

To address high cache coherence overhead and poor scalability in disaggregated memory (DM) architectures, this paper proposes DiFache, a decentralized cache framework. Its core contributions are: (1) aligning the coherence model with memory nodes—not CPU caches—by leveraging the inherent serialization of application-level remote memory accesses; (2) designing a coordinator-free invalidation protocol enabling independent operation of remote compute nodes; and (3) introducing a fine-grained, adaptive object caching policy that dynamically adjusts cache granularity based on observed read-to-write ratios. DiFache tightly integrates RDMA primitives with DM semantics. Evaluated on 54 real-world Twitter trace workloads, it achieves an average speedup of 5.53× (up to 10.83×). When integrated into two production DM applications, it improves peak throughput by 7.94× and 2.19×, respectively.

Technology Category

• Data Mining & Knowledge Management: Scalability, Parallel & Distributed Systems
• Search and Optimization: Distributed Search
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Data management and stream processing for Web, mobile and wireless applications
• Security and Privacy: Large-scale security measurements
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

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$ imes$ (5.53$ imes$ on average). By integrating DiFache, the peak throughput of two real-world DM applications increases by 7.94$ imes$ and 2.19$ imes$, respectively.

Problem

• Reduces access overhead in disaggregated memory via CN-side caching
• Eliminates centralized cache coherence maintenance for scalability
• Optimizes caching for varying read-write ratios adaptively

Innovation

• Decentralized coherence for DM caching
• Decentralized invalidation on remote CNs
• Fine-grained adaptive caching scheme