Octopus: Scalable Low-Cost CXL Memory Pooling

2025-01-15

AI Summary

Existing CXL memory sharing relies on expensive switches, suffers from poor topological flexibility, and incurs high deployment costs. To address these limitations, this paper proposes a low-cost, highly scalable CXL memory pooling architecture. Our approach introduces: (1) Octopus—a novel lightweight, heterogeneous, and asymmetric topology—enabling cross-server memory sharing using only off-the-shelf, small-form-factor CXL Type-3 devices; and (2) a hardware-agnostic, lightweight memory virtualization management mechanism built upon the CXL 3.0 specification, eliminating the need for proprietary switching chips. Experimental evaluation demonstrates that, at three times the host scale, per-host cost is reduced by 17%, while memory utilization and cluster scalability are significantly improved. This work is the first to empirically validate the feasibility of low-latency, high-density CXL memory pooling under stringent cost constraints, thereby advancing the democratization and practical deployment of CXL technology.

Technology Category

• Machine Learning: Hardware-aware ML
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization
• Search and Optimization: Distributed Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Virtualization and resource management in Web systems and infrastructures
• Economics, Online Markets and Human Computation: The sharing economy
• Security and Privacy: Large-scale security measurements

Abstract

Compute Express Link (CXL) is widely-supported interconnect standard that promises to enable memory disaggregation in data centers. CXL allows for memory pooling, which can be used to create a shared memory space across multiple servers. However, CXL does not specify how to actually build a memory pool. Existing proposals for CXL memory pools are expensive, as they require CXL switches or large multi-headed devices. In this paper, we propose a new design for CXL memory pools that is cost-effective. We call these designs Octopus topologies. Our design uses small CXL devices that can be made cheaply and offer fast access latencies. Specifically, we propose asymmetric CXL topologies where hosts connect to different sets of CXL devices. This enables pooling and sharing memory across multiple hosts even as each individual CXL device is only connected to a small number of hosts. Importantly, this uses hardware that is readily available today. We also show the trade-off in terms of CXL pod size and cost overhead per host. Octopus improves the Pareto frontier defined by prior policies, e.g., offering to connect 3x as many hosts at 17% lower cost per host.

Problem

• CXL Interconnect
• Memory Sharing
• Cost-effective Topology

Innovation

• Octopus Topology
• CXL Standard
• Cost-effective Memory Sharing