DockerCon thoughts..

So I had the great pleasure to attend DockerCon in San Fran this year, and what a DockerCon it was. It seems every year the number of "its my first ever DockerCon" attendees keeps increasing, which is a great sign for the technology's adoption.

This year i decided to host a number of hallway tracks, and also ran a beer and pizza meetup on the monday night prior to DockerCon. Through these, i met a large number of awesome Portainer users, shared a demo of our upcoming RBAC extension, and gathered feedback on our product and how its used. Expect to see another blog from me on our future strategy soon.

I also had the privilege of meeting Steve Singh and his right-hand man, David Messina, and was able to share the Portainer vision, and how we can work together with Docker to further adoption of container technologies.

Finally, I volunteered myself to be a Teaching Assistant/Aide for Bret Fisher and his Docker Swarm paid workshops on the Tue and Wed. These events saw a great turnout, with a full house both days. Got to say, running a "follow along" workshop with 100 people following along is extreme, keeping everyone going at the same speed is quite a challenge (but fun). I enjoyed being able to run around and help those that got stuck or needing a little more explanation. Bret is a wizard teacher though.

Anyway, DockerCon..

It has become increasing apparent that the future Docker see "in the tea leaves" is one where organisations do not "buy" off the shelf containerised applications from ISV's, but one in which they either build (from source), or assemble (stitch together a bunch of open source components/frameworks). This is evident in Docker's focus to push more and more functionality into Docker Desktop (Enterprise Edition) that aids the developer to build, publish, and support their application bundle in production.

The application builder functionality of Docker Desktop enterprise is particularly neat, as you simply select the components you want to build your app around (DB, Web Server, framework etc) and the builder stitches it all together for you as a service, and then lets you deploy that service locally (on swarm or kubernetes), or on a centrally managed Docker Enterprise UCP environment via Docker Trusted Registry (again, swarm or kubernetes). Clearly docker see Docker Desktop as the way developers should interact with Docker Enterprise, rather than using the UCP/DTR web interfaces (in fact, all things being equal, developers should not need to login to UCP/DTR if they construct their apps in the right way - ie can just destroy and redeploy if any issue - no need to troubleshoot).

The other cool feature of Docker Desktop is that you can now build application environments based on multi-arch images, so your "bundle" can automatically deploy on any architecture (ARM/x86/PPC) that your multi-arch images support. The actual execution of docker containers on any architecture has been a thing forever, and so too has multi-arch images (one image that supports multiple architectures), however Docker have now made is so simple to create a stack/service, package it up and distribute it for use everywhere.

Docker also announced their new "Docker Enterprise as a Service" offering, which is a subscribed managed solution of Docker Enterprise, delivered by authorised docker partners, and is deployed, licensed, and managed by the partner, with the customer simply "using/consuming" the service. This is an interesting offering as it seems customers must be demanding this managed solution from Docker. I wonder if Docker Enterprise still requires a high enough degree of operational management so as to warrant paying someone else to manage it for you. Clearly so..

Finally, Docker announced their Docker Kubernetes Service (and also the Docker Swarm Service), which aims to offer a consistent orchestration experience (versions etc) between Docker Desktop and Docker Enterprise.

One notable theme this year was the resurgence of Docker Swarm.. there were a number of sessions specifically on Swarm, even one on a customer that moved from Swarm to Kubernetes, realised it was an operational nightmare, so moved back to Swarm again. It was also awesome to attend the Swarm futures session, where we were able to hear first hand from the Swarm Dev team whats upcoming, and let me tell you, there is some cool stuff coming. It was also intrieging to hear that the vast majority of Docker Enterprise customers actually still use Swarm, and at massive scale. Who says swarm is dead.. its far from it.

Anyway from Portainer's perspective, we are confident that remaining aligned with Swarm is the right move (although we are investigating how to make Kubernetes as simple to use in our UI as Swarm, which is no easy task!). We also have a slight difference of opinion to Docker; we believe that a large proportion of organisations are not able to invest (either time or money) to build large teams of high trained developers tasked to create innovation, and instead would look to "buy" instant innovation made available from third party specialist application vendors, who still need to be able to deploy their apps on Docker. For this reason we will continue to provide both a fully featured API for Developers to control Portainer, as well as our more traditional IT Op's style "opinionated" interface to enable any IT team to manage Docker.

So, all in all, and awesome DockerCon, great to see Docker growing, great to see products evolving, and great to see the adoption continuing.

Lets get dockerizing..


Leave a comment!

All fields marked with an asterisk* are required.