Summary

This Supplement identifies the technical considerations for designing networks in cloud infrastructure to support key workloads such as machine learning (ML), high-performance computing (HPC), and distributed storage. It analyses the unique network traffic characteristics generated by these workloads. This Supplement establishes a technical framework covering scalable non-blocking networks, high-performance data transport, and the abstraction of advanced network capabilities to guide cloud service providers and network architects in building high-performance, efficient, and reliable network infrastructures.