AI for Network Engineers: Understanding Flow, Flowlet, and Packet-Based LB

For a latest reference on AI and machine learning for network engineer please check this book by Javier Antich [1].<p>Please also check the review here [2]. For what it&#x27;s worth, the book is listed in the &quot;10 Books Every Network Engineer Should Read&quot; [3].<p>[1] Machine Learning for Network and Cloud Engineers: Get ready for the next Era of Network Automation:<p><a href="https:&#x2F;&#x2F;www.goodreads.com&#x2F;book&#x2F;show&#x2F;101180344-machine-learning-for-network-and-cloud-engineers" rel="nofollow">https:&#x2F;&#x2F;www.goodreads.com&#x2F;book&#x2F;show&#x2F;101180344-machine-learni...</a><p>[2] MUST READ: Machine Learning for Network and Cloud Engineers:<p><a href="https:&#x2F;&#x2F;blog.ipspace.net&#x2F;2023&#x2F;02&#x2F;machine-learning-network-cloud&#x2F;" rel="nofollow">https:&#x2F;&#x2F;blog.ipspace.net&#x2F;2023&#x2F;02&#x2F;machine-learning-network-cl...</a><p>[3] 10 Books Every Network Engineer Should Read:<p><a href="https:&#x2F;&#x2F;networkphil.com&#x2F;2024&#x2F;05&#x2F;21&#x2F;10-books-every-network-engineer-should-read&#x2F;" rel="nofollow">https:&#x2F;&#x2F;networkphil.com&#x2F;2024&#x2F;05&#x2F;21&#x2F;10-books-every-network-en...</a>

This is not &quot;AI for network engineers&quot; but rather &quot;Network engineering for AI datacenters&quot;. I was expecting to read that a small neural network could be used to direct traffic.

Far more practical than I expected. I particularly enjoyed the detailed diagrams.

&quot;Though BGP supports the traditional Flow-based Layer 3 Equal Cost Multi-Pathing (ECMP) traffic load balancing method, it is not the best fit for a RoCEv2-based AI backend network. This is because GPU-to-GPU communication creates massive elephant flows, which RDMA-capable NICs transmit at line rate. These flows can easily cause congestion in the backend network.&quot;