aiqbench takeChips & Infra

This article details how Amazon Web Services has transitioned from traditional "fat-tree" topologies to scalable "flat" networks using quasi-random graph designs and new passive optical components called ShuffleBoxes. By reducing router overhead and improving routing resilience, this shift represents a significant architectural evolution in the infrastructure supporting large-scale AI model training and inference workloads.

Sources coveringAmazon Science, The Verge, Latent Space, SiliconANGLE

What this isA model's reading of the article and related coverage, written for aiqbench. Not the publisher's words.

How flat is replacing fat in AWS data center networks

Amazon ScienceMay 28, 20267 min read

“Quasi-random” network topologies and new passive optical components called ShuffleBoxes make more-efficient flat networks as practical as traditional “fat-tree” networks.

Routing in today’s data centers is usually governed by a data structure called a “fat tree”, which is similar to a corporate organizational chart, with nodes in each layer connecting to multiple nodes in the layer below. Here, however, the nodes of the bottom layer represent routers that want to send messages to each other, and the layers above them contain extra routers that simplify the routing procedure. A message sent by one bottom-layer router climbs the tree until it reaches the branch that leads to the destination router, and then it is sent down. A fat tree. Each node represents a router, and each router has four ports. Nodes T1 – T12 reserve two ports each for connecting to servers. This design is easy to implement but inefficient: the…

Excerpt shown under the publisher's RSS terms · 130 words · full text and figures are on the publisher's site

Read at Amazon Science ↗Ask about this article