RabbitMQ-less OpenStack Cluster - Supporting 13,000 Hypervisors in One Region

Learn about an innovative solution for large-scale OpenStack deployments in this 28-minute technical talk that addresses the challenges of using RabbitMQ as a messaging backend. Discover how to overcome common bottlenecks in clusters with over 1,000 hypervisors through the implementation of a new oslo.messaging driver. Explore the architecture and technical details of a broker-less, one-to-one direct RPC call system operating over HTTP protocol, successfully deployed in a production environment managing 13,000 hypervisors. Gain valuable insights from two years of operational experience with this RabbitMQ-less solution, including common troubleshooting techniques and practical lessons learned from real-world implementation.