Problem
Research questions and friction points this paper is trying to address.
Extreme-Scale Interconnection Networks
Leaf-Spine Topology
Scalability
Throughput
Data Center Networks
Extreme-Scale Interconnection Networks
🤖 AI Summary
This work proposes Multipass Random Leaf-Spine (MRLS), a novel interconnect architecture designed to overcome the limitations of Fat-Tree topologies in hyperscale data centers with respect to scalability, throughput, and cost efficiency. MRLS integrates the structural advantages of Orthogonal Fat-Tree and Random Folded Clos, augmented with non-shortest-path routing strategies, to achieve significantly enhanced network performance while preserving high flexibility. Experimental evaluation demonstrates that under an all-to-all communication pattern involving 100,000 endpoints, MRLS delivers 50% higher throughput than conventional Fat-Tree and doubles the throughput of Dragonfly, thereby exhibiting superior scalability and cost-effectiveness for large-scale deployments.
📝 Abstract
Extreme-scale data centers are the backbone of next-generation computing, enabling breakthroughs in science, artificial intelligence, and global innovation through unprecedented processing power and scalability. This work examines leaf-spine network topologies that offer extreme scalability--connecting a vast number of endpoints--while delivering strong performance at low cost. It takes as a starting point two alternatives to the widely used Fat-Tree topology: the Orthogonal Fat-Tree and the Random Folded Clos. The resulting Multipass Random Leaf-Spine (MRLS) networks inherit their advantages and surpass Fat-Trees in both throughput and flexibility. To fully leverage the topological properties of these networks, various non-minimal routing strategies are considered. An exhaustive evaluation using an interconnection network simulator provides insight into the trade-offs and scalability of these topologies under realistic conditions, positioning them as a promising solution for extreme-scale systems. The MRLS achieves a 50% speedup against a Fat-Tree for an All2All collective comprising 100k endpoints, and 100% against Dragonfly networks for the same collective.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Multipass Random Leaf-Spine
extreme-scale interconnection
non-minimal routing
network topology
All2All collective
🔎 Similar Papers
No similar papers found.
