Case Study: a major telecom company achieves 40% TCO savings with Aviz Networks' vendor-agnostic AI fabric manager

Implementing a Vendor-Agnostic AI Fabric Manager for Enhanced GPU-as-a-Service Operations

The major telecom company faced significant challenges with its single-vendor AI infrastructure, which caused innovation stagnation, high costs, and competitive risk. To build a future-proof GPU-as-a-service cloud, they needed to overcome complex issues with interoperability, monitoring, and managing a multi-vendor RoCE-based AI fabric. They partnered with Aviz Networks to address these strategic problems.

Aviz Networks implemented its ONES and SONiC solutions to create a vendor-agnostic AI network fabric. The solution included configuring RoCEv2, optimizing 400G server NICs, and designing a large-scale fat tree network topology. This provided the customer with 40% TCO savings, ensured SLA compliance with 24/7 monitoring, and enabled efficient, high-performance GPU-as-a-service operations.