DPC: A Distributed Page Cache over CXL

TL;DR

DPC is a novel distributed page cache leveraging CXL 3.0 memory semantics to reduce redundancy and coherence overhead, significantly improving performance in distributed file systems.

Contribution

This paper introduces DPC, a distributed page cache that enforces a single-copy invariant across a cluster using CXL 3.0, enhancing efficiency and scalability.

Findings

DPC achieves up to 12.4X speedup on real workloads.

DPC reduces data redundancy by maintaining a single owner per page.

DPC demonstrates significant performance improvements over traditional caches.

Abstract

Modern distributed file systems rely on uncoordinated, per node page caches that replicate hot data locally across the cluster. While ensuring fast local access, this architecture underutilizes aggregate cluster DRAM capacity through massive data redundancy and incurs prohibitive coherence overhead via heavyweight, lock-based protocols. In this paper, we focus on the design of a distributed page cache that treats the entire cluster's main memory as a single cache budget while preserving standard file-system interfaces and semantics. We present Distributed Page Cache (DPC), an OS-level, distributed page cache built on top of Compute Express Link (CXL) 3.0 memory semantics. DPC enforces a single-copy invariant at page granularity: each file page has exactly one owner node holding the sole resident DRAM copy, and other nodes access it via CXL-based remote mappings rather than creating replicas of the page. DPC is implemented end-to-end on a CXL-based emulation framework that models multi-host CXL 3.0 memory fabrics, enabling detailed evaluation in the absence of widespread hardware. Across real-world and representative data-sharing workloads, DPC delivers speedups of up to 12.4X, with a geometric-mean speedup of 5.6X.