InfiniBand has been the gold standard for AI clusters, delivering ultra-low latency and high throughput. However, Ethernet—specifically RoCEv2 (RDMA over Converged Ethernet v2)—is rapidly emerging as a competitive alternative, offering greater flexibility, cost efficiency, and broader industry adoption.
This session, led by experts from World Wide Technology, will provide an in-depth comparison of InfiniBand vs. RoCEv2, exploring the strengths, limitations, and considerations for AI workloads.
Attendees will gain insights into:
Whether you’re designing a new AI infrastructure or optimizing existing AI networking, this session will help you determine the right approach to building a high-performance, scalable, and cost-effective AI networking strategy.
I love to solve data center networking challenges. The last five years has changed my focus significantly to Cisco ACI and VMware NSX. In many cases this includes a discussion of how to integrate the two solutions together.
My second passion is Motorsports. Over the course of the last 30 years I have been heavily involved working with F1, Indycar, IMSA, many other sanctioning bodies and their events.
Register now and receive exclusive access to ONUG content and updates
Register Here