DockerCon Day 1 – what you’re missing


Austin you are so chill. It is way too quiet here for the onslaught of nerds attending DockerCon17. If you are missing it, this is what you are now behind on.

Docker is open sourcing its way into maturity

Borrowing a page from the RedHat playbook. Docker is committed to the community and will continue the trend of open-sourcing key core components of Docker to the masses. By doing so they hope to be the name, and technology at the heart of any container implementation. And they’re on to something. First the third party vendors stand to benefit huge by having even tighter integrations with container technology. And potential competitors won’t re-invent the wheel. Which keeps them at arm’s length from the Docker talented engineering team at all times. But how will they make money! The two big announcements on day one were LinuxKit and Moby Project.

LinuxKit: I’m still trying to wrap my brain around this. Not what it is, but why it is. I guess I did not realize the demand for tiny hardened containers was so high. For your typical application I don’t see LinuxKit being terribly beneficial. Actually probably just more work. But for IoT, like medical devices etc this is very cool. Given the partner lineup (ARM, Intel) that seems to be where it is going. And I’m wondering why such a niche announcement at the keynote. What is the backstory? Multi-platform required the building of a Linux subsystem. Maybe it’s as simple as that.

I do see that LinuxKit further supports the demand for immutable infrastructure. And the idea that everything should be a container. The technology is great fast as hell, shortest possible boot. But of the 12 vendors on the show floor I talked to, it does not seem like it will be incorporated widely into offerings.

Moby Project:And Moby project seems to tie it all together. It is an umbrella for all the open source shenanigans. A great way to centralize the community. I suspect it will be used to gauge what is needed to build out an even more robust container ecosystem. And maybe get out in front of those brave techno-soldiers who actually dig into the code bases.

What I struggle with is the message seeks a bit conflicting. Solomon started the keynote by saying developers need to drop complexity where it can be. Stop being tinkers just for fun of it. But at the same time opening up some rather complex technology to a world that otherwise should not even need to know about it.

It started with  containerd then notary and now LinuxKit. Every time they open source something my internal nerd tells me “oh you are totally going to play with codebase tonight”. And i never do. Technically it’s very cool, and great to feel the love from a key vendor in the space. But I think that is where most of the value stops. Besides the niche awesome application, most of the third party tool vendors I spoke to have not leveraged the projects to improve their offering. With the exception of Twistlock who uses notary, and has also made contributions of their own to Docker authorization. It’s hard to see the outcome of these projects.

So for now these announcements i’m going to chalk up to some serious nerd cred. There were a few other smaller announcements about enhancements to the product which seem very cool.

In response to “Images are too big!” – Docker added multi-stage builds. Separate build image from production environment. So you don’t have to ship whole containers. I’m not sure how it will impact security? But I do know that it will help with the “move your monolith” adoption of containers. And also speed up any deployment by only shipping essentially the diff.

In response to “It’s hard to move to cloud” – Docker is creating greater parity and faster deployments with Desktop-to-cloud. You can automate your connect to your cloud through tunnel. Which brings the cloud to you to the developers. But it requires Swarm.

Diago Monica the lead security engineer talked a lot about security features as well. But I did not understand a word. I really love that they are talking security, and this is going to really help mature deployments.
This years announcements are on par with last years. And tomorrow I think the tone will change and explain more about the business. To my pleasant surprise, I did not see a single reference to Hub downloads. Which in previous years was the one key metric used to define success. But the problem is it was only an adoption from devs in sandbox us case.

Chris Riley is a technologist and DevOps advocate for @Splunk who has spent 12 years helping organizations transition from traditional development practices to a modern set of culture, processes and tooling.


Click on a tab to select how you'd like to leave your comment

Leave a Comment

Your email address will not be published.

Skip to toolbar