-
Notifications
You must be signed in to change notification settings - Fork 0
Production Containers Workgroup
The Production Containers Workgroup is a public forum to discuss what it takes to run containers (vs. VM's) in production. By in production, we mean being able to achieve required aspects of automatic recovery (fault tolerance), high availability, elastic and web scale, operational visibility, compliance, security and continuous delivery. We realize that containers have great value beyond these production scenarios (for example the improvements possible in continuous integration and developer productivity in production like environments) but for this working group we wanted to look at the production operation requirements specifically.
While this workgroup will consider containers using various technologies, this group is not a group to help you decide on what technologies to use. It is our hope that this workgroup will drive all container related technology forward in production readiness as well as realize points of collaboration in related container technologies.
At Dockercon 2014, we heard of many users and implementors working with containers in production through Docker but wanted to create a forum where we could talk about the work in this area. Also, we realized that the majority of the usage across the world was more focused on development time experiences with the assumption that production usage of containers will be just as operationally possible as virtual machines on clouds.
In July 2014, IBM, Mesosphere, and Google met to talk about an approach to help drive this discussion forward. We decided on the following aspects:
-
Name the effort "Production Containers". We considered many other names, but found the name to be the most meaningful terse name we could come up with.
-
Be as open on the discussion as possible. This means we'd share our experiments through source on github, documentation via github wikis and discussions through a Google Group.
-
Settle on a few applications to start with that show a representative class of applications we'd want to run in production. We settled on an on-line interactive multi-tier applications suite, a batch off-line processing of large data where final response isn't as time sensitive, and an off-line continuous build system that was more time sensitive.
-
We discussed some high level areas that we'd want to test in operational/production scenarios. These scenarios will be discussed on this wiki with results available through community testing.
-
We also talked about the need to have validation of users or the workloads, scenarios, and results. We ask anyone interested in joining the working group to note their interest in the Google group.
IBM
Mesosphere