From Datacenter to AI Center, building the networks that build AI

Описание: Abstract

As Artificial Intelligence (AI) workloads move from experimental phases to massive production scale, the underlying data center network must evolve to meet unprecedented demands for bandwidth, low latency, and lossless communication. This presentation provides a foundational roadmap for network engineers transitioning from traditional general-compute architectures to specialized AI networking environments.

Key technical pillars covered include:
Architecture & Topologies
Congestion Control
Load Balancing
Segmentation
Operations & Visibility

Tyler Conrad:
Tyler Conrad is a Systems Engineering Tech-Lead for Arista based out of Austin, TX, where he focuses on the intersection of AI, routing, and datacenter network design. With a 15-year career rooted in semiconductor manufacturing and private cloud, he has spent over a decade as a "hat collector" across the networking and security landscape. In 2020 he had a sudden influx of free time and started dedicating his expertise to the network infrastructure requirements of AI. A former expat to Mexico and a speaker of several languages (with varying degrees of success), Tyler balances his technical rigor with high-adrenaline hobbies like rock climbing, mountain biking, and amateur electrical work.

https://nanog.org/events/nanog-96/con...