Toward 10,000 Containers on OpenStack
Kubernetes, Swarm and Mesos have been shown to scale to hundreds or thousands of nodes and tens of thousands of containers, handling millions of requests per second. On OpenStack, they can be easily deployed and managed by Magnum, but how do they scale? Building the correct infrastructure plays a critical role in the performance and scalability for the containers, and this needs to be implemented in Magnum.
Working toward this goal, we have built a collection of Rally plugins to track the performance and scalability of both Magnum and the Container Orchestration Engines (Kubernetes, Swarms, Mesos). We have been running the benchmarks on large OpenStack environments to collect measurements and study the behavior of the system. In this talk, we will present our findings along with some of the best practices we found for operating large container environments.
