Resilient AI Supercomputer Networking using MRC and SRv6

TL;DR

This paper presents a resilient networking approach for AI supercomputers combining MRC, multi-plane Clos topologies, and SRv6 source-routing to improve fault tolerance and performance at large scale.

Contribution

It introduces a novel RDMA-based transport protocol, MRC, integrated with SRv6 static routing and multi-plane topologies for large-scale AI training clusters.

Findings

MRC effectively load-balances and sprays traffic across multiple paths.

The topology and routing design enhances fault tolerance and reduces tail latency.

Successful deployment in OpenAI and Microsoft clusters demonstrates practical benefits.

Abstract

Tail latency dominates the performance of synchronous pretraining jobs when running at very large scales. We describe a three-pronged approach: (1) a new RDMA-based transport protocol, MRC, sprays across many paths and actively load-balances between them, eliminating the issue of flow collisions (2) the use of multi-plane Clos topologies to get the benefits of high switch radix and redundancy, allowing training clusters well over 100K GPUs to be built as two-tier topologies while increasing physical redundancy, and (3) the use of static source-routing using SRv6 to allow MRC the freedom to bypass failures by itself. We describe our experiences running MRC and static SRv6 routing in production in OpenAI and Microsoft's largest training clusters, where it has been used to train the latest frontier models. We demonstrate how MRC allows AI training jobs to ride out many network failures that previously would have interrupted training.