Case Study: a leading cloud service provider achieves cost-effective GPU cloud performance with Aviz Networks ONES

Leveraging RoCE for Cost-Effective GPU Cloud Services

A leading cloud service provider faced the challenge of enhancing the RDMA over Converged Ethernet (RoCE) performance for its GPU services without succumbing to the high costs and vendor lock-in associated with proprietary solutions from vendors like NVIDIA and Arista. They sought a cost-effective, high-performance solution to support AI and machine learning workloads.

The provider implemented the Aviz Open Network Enterprise Suite (ONES) from Aviz Networks. This vendor-agnostic platform provided an open, scalable architecture with expertise in SONiC and an AI fabric design. The solution enabled the customer to launch their first GPU cluster in just three months, achieve 40% savings in total cost of ownership (TCO), and gain a competitive edge by avoiding vendor lock-in.