The test demonstrates Nokia's compatibility and commitment to integrating the capabilities of the Ultra Ethernet Consortium (UEC) specification 1.0 into its portfolio of HPC/AI-ready networks, and reinforces its leadership in building extreme-scale, lossless, low-latency networks for modern data centers built for the AI era.
AI is driving a fundamental rethinking of data center operations. Real-time training and inference demand ultra-low latency, while even minimal packet loss can derail training jobs and delay their completion. Combined with enormous bandwidth requirements, these pressures are pushing existing data center networks to their limits.
The Ultra Ethernet Consortium (UEC) Specification 1.0 defines this new UET layer to address the complex set of challenges created by HPC/AI. Given the widespread adoption of Ethernet in most data networks worldwide, this is an attractive protocol for modernizing AI data center networks with new standards, best practices, and architectures that support AI and HPC workloads. As an active member of the UEC, Nokia is committed to leveraging UET as a highly compatible, cost-effective, and interoperable component of AI and HPC application stacks.
The procedure tested 800 GE interfaces with UET traffic flows generated by Keysight's AresONE 800GE-8P-QDD-M test platform across a network spanning all variants of the Nokia 7220 IXR-H5 and Nokia 7250 IXR-10e platforms. These Nokia switches ran the SR Linux network operating system (NOS).
As Nokia integrates the capabilities of the UEC 1.0 specification into its Nokia data center fabric solution for HPC/AI networks, global customers now have immediate commercial access to a UEC-compliant, AI-ready portfolio with support for Direct Remote Memory Access over Converged Ethernet (RoCEv2) and Quantified Data Center Congestion Notification (DCQCN). Current RoCEv2 support was also tested on parallel 800 GE flows in the same testbed, demonstrating its coexistence, openness, and flexibility.
