KubeCon Europe: Day 2 Summary and Retrospective

Following up on the Kubernetes 1.2 news we reported in our day 1 summary post, v1.2.0-beta.0 and v1.2.0-beta.1 have been released on Github. There are still a few priority 0 issues, but I anticipate a release in the next couple of weeks.

Once again, this post won't cover all the talks as KubeCon EU was split into 4 different tracks! I've been assured that every talk was recorded, so when available I'll update this post with links to the videos as and when they're available.

Day 2 Summary

The day 2 Keynote, "Pushing Kubernetes Forward" was delivered by Brandon Philips - CTO at CoreOS. He showed off the astounding progress Kubernetes has made in the last few months due to community contributions (you can see last year's progress illustrated by us here!). If you feel like kubernetes is missing something, you should fork it on Github, and participate in the 30+ pull requests it gets per day.

Michael Ward, Principal Systems Architect at Pearson, talked about "Orchestrating an enterprise": how containers and Kubernetes can revolutionise build pipelines in large enterprises. By getting everyone to use the same base images, updates are in small, security auditable layers - meaning each build can be pre-approved for deployment. Build servers become cattle (Ed - bingo!), and the build, test and deploy process no longer blocks other teams, saving a ridiculous amount of everyone's time.

Maciej Szulik from Red Hat talked about an under-utilised feature of Kubernetes - Batch Jobs. Kubernetes has had experimental support for them for quite a while, but they will be added to the main API in v1.2. Maciej put forward some use cases for batch jobs, then showed how to use the Kubernetes Job object by calculating pi. We also look forward to seeing Scheduled Jobs make their way into Kubernetes soon.

Meanwhile, Matthew Garrett, Principal Security Engineer at CoreOS talked about securing your whole cloud platform. How do you know the container you're running is the one you originally pushed? He argues the only way to know for sure that your container host hasn't been compromised is to use trusted computing - which verifies the operating system the computer has booted.

After lunch, Erin Boyd, also from Red Hat then gave an intro to Kubernetes storage options. Various volume plugins were discussed, and we were shown how to add volumes to Pod templates. Discussion continued into the area of managing volume security with OpenShift.

Next, our turn to shine. David O'Dwyer - founder of LiveWyer (that's us!), and Christoph Held - Cloud Architect at Fujitsu shared the stage to show off the upcoming Kubernetes Dashboard project. The Kubernetes Dashboard is a general purpose cluster UI, replacing the old Kube-dash / Kube-UI. They looked at what various users would want from a UI, and explored areas of upcoming development. (Ed - Please submit your thoughts on the Kubernetes Dashboard here! </plug>)

Finally, Ian Lewis, Developer Advocate at Google, talked about Kubernetes' rich API and how to use it. He showed off some very cool demos that used physical hardware to manipulate Kubernetes, and discussed the possibilities for what apps can do with the API.

Retrospective

After another great KubeCon (and post-KubeCon gathering) we can look back at how far Kubernetes has progressed in the last 6 months. Most of the features we were sorely missing in Production, such as configuration management, Ingress and ConfigMap and practical Operations resources - Deployment, are being solved in version 1.2.

After seeing the roadmap for release 1.2 and release 1.3 we are even more excited for the future of Kubernetes as a practical, scalable, portable and most importantly - open system for organisations to build solutions around.

Well done to the KubeCon Team - See you again later in the year!