Abstract
Mesosphere used to migrate its applications to containers to isolate the applications and make them easily shippable. But now the company is struggling to run stateful applications to make them scalable and fault-tolerant. And it runs its big data and fast data frameworks in separate clusters, because it is not super easy to integrate them into the current container stack. This means that Mesophere’s cluster utilization is not optimal and scaling affects comprehensive containers and big data frameworks. This session looks at current container usage, shows the actual problematics, and discusses how DC/OS and Apache Mesos deliver “Container 2.0.”
