Automated Congestion Management in AI Data Centers - Network Optimization with DCQCN

Watch a technical presentation from Cloud Field Day 20 exploring how Juniper Networks addresses automated congestion management in AI/ML data center fabrics. Dive into the complexities of DCQCN congestion management protocol and learn how Juniper Apstra's innovative solution automates network optimization. Follow along as Rajagopalan Subrahmanian and Vikram Singh demonstrate their DCQCN AutoTune application, which leverages continuous monitoring and closed-loop automation to dynamically adjust network configurations. Understand the interplay between Priority Flow Control (PFC) and Explicit Congestion Notification (ECN) in managing network congestion, and discover how automated systems can maintain optimal throughput while balancing packet loss. Explore real-world applications through lab simulations that showcase the system's ability to detect congestion and implement configuration changes using Terraform across network fabrics. Recorded in Sunnyvale, California, this 24-minute presentation offers valuable insights into modern solutions for managing complex AI data center networks.