Container Orchestration Engines such as Kubernetes and Swarm provide full management for containers, however they assume that the cluster infrastructure is provided by someone else. OpenStack provides the full infrastructure and Magnum integrates these resources for the Container Orchestration Engine, but the resource management is left to the users. Mesos on the other hand provides advanced, highly scalable resource management.
Recent development allows Kubernetes and Swarm to run on Mesos as frameworks. This makes it now possible to bring the 3 layers together to achieve a new level of integration and automation, yielding many benefits: scaling at both the container and the node level, fine grained resource sharing among clusters, etc.
In this talk, we will take a close look at running Kubernetes and Swarm workloads on Mesos, hosted on OpenStack. We will show the necessary plumbing for the integration, and discuss potential issues in providing networking and storage through Mesos.
The audience will gain an understanding on how well Mesos and OpenStack complement each other to host disparate container workloads.
We will identify best practices as well as gaps that may drive new development in the respective communities.