Or how Open Source ‘community building’ is really tricky, and what it means for the future of the container ecosystem.
Last week I attended my very first DockerCon in Barcelona, Spain. Back in September, I swallowed my sticker shock and bought a ticket. I then spent the next several weeks anticipating my journey. I mean, what’s not to love about a container nerd-fest set in an awesome convention center near the beach of a classic and food-friendly Mediterranean city?
Aside from the obvious attraction of the event, I was also in dire need of an excuse to reconnect with what is coming next. As Packet’s CEO, a big part of my value is leading the strategic vision for our company. Essentially: how can our bare metal compute platform best serve the needs of modern workloads?
With the container world’s breakneck speed of evolution, I try to keep my finger on the pulse through constant personal engagement with customers, partner and competitors. But honestly, nothing beats sitting down with all the smartest minds in person and it’s REALLY hard to do that one by one. As I checked around my network to see who was attending, it became clear that DockerCon was going to be a great intelligence gathering opportunity!
The conference was Sunday night through Tuesday evening. I booked my tickets, arranged for a stop over in London, grabbed an AirBNB with another bare metal nut (@zehicle) and organized a Junto dinner with the best minds I could find.
I was set! And boy did I have questions to answer.
Questions on the Brain
As I made my way to Barcelona, a few key topics started formulating in my head: things about which I needed to learn more or for which my opinions weren’t quite strong enough. Some of these were stimulated by hints I had received from Dave Laube, our head of platform engineering (who had just attended KubeCon in early November) as well as the ContainerDays summit I had attended in NYC.
The list looked like this:
- What clues could I find about Docker’s business model? Where and when were they most likely to turn their massive community and adoption wave into revenue?
- What was the status of enterprise adoption (in production) of container tech? Which vendors, platforms or tools were getting the most traction in those production environments?
- What was going on with security? What was the state of IPv6 adoption? How about storage drivers? Some feature detail stuff.
- How strong / satisfied was the community? What was the pulse of the Docker community and vendor ecosystem?
- What about Kubernetes? How would Kubernetes fit into the Docker world?
Within a few hours of the opening keynote, I could see that the big topics to unravel would be related to Kubernetes and the Docker community, with the Docker business model coming in right behind those (and maybe the cause of all the drama in the first place). Here are some thoughts about each of these...
It’s obvious that Docker has been successful at building a community -- the conference was packed in Barcelona. Docker “whale” tshirts were on nearly every attendee and there was an engaged and diverse group of contributors to the open source project.
But after living through original open source communities around Linux, the Apache projects, Mozilla, OpenStack and Android, the Docker one smells a lot more consumer. Basically there are a large number of people involved, but many of them are not writing core software - simply living and liking what Docker stands for. One could say that open source has finally become cool.
Another part of the community was present as well: the vendors building businesses around the Docker / container ecosystem. Build tools, deployment platforms, on-premise container suites, etc. The vibe within this community (i.e. the actual sponsors of the conference) was much more muted.
Most vendors I spoke with were either openly upset or guarded about two things: Docker’s commercial objectives and a top-down approach to its ecosystem. With the recent acquisition of Tutum, the announcement of Docker Control Plane and the various signals around Docker networking and storage layers, the vendor ecosystem is rightfully on edge about Docker’s long term intentions. In short: will Docker openly compete with (or roadmap away) key members of the vendor ecosystem?
As for Kubernetes, the open-source (apache licensed) container cluster managing project from Google, it was mindfully absent in the billboards and handouts of the vendors, or in the presentations on stage and in the breakout rooms. In fact, pretty much everything at DockerCon was 100% Docker and had been wiped clean of competitive or alternative tools, ecosystems or debate.
I came away with a distinct sense that Docker, Inc fears Kubernetes in some way, while the vendor ecosystem was openly excited and significantly more embracing of the project than even a few months ago. I believe this has little to do with tech. Is Docker Swarm or Kubernetes better, easier, faster or cooler? Who knows! But what is clear is that the economic intentions of Alphabet, Inc are not about monetizing an open-source container manager, but about growing a healthy ecosystem that thinks and acts positively towards Google-style tooling (and Google Container Engine!).
Docker has all the incumbent advantages in the container world including massive brand recognition, developer mindshare, a huge ecosystem headstart and plenty of funding. However, I’d watch for Kubernetes and its flank (Tectonic, Kismatic, Deis, Digital Rebar, etc) over the next few months. My hunch is you’ll see some Docker-only tools and vendor partners start supporting Kubernetes as well or, at the very least, begin hedging their bets in case Docker, Inc moves into their world.
Open Source Business Models
This brings me to my other big take-away from my trip, basically around open source business models. DockerCon is living proof that giving away free software to developers, and betting on developers driving purchasing power down the road, is a business model that is pretty well accepted. Heck, even Basecamp is in on the game (joke)... But it’s important to note that very few companies have built successful complex opensource ecosystems.
People will point to select open source projects making gobs of money (RedHat and their brethren) as proof that the formula works, but I asked the members of the Junto dinner (many of whom work for open-source business model companies) what was the last successful ecosystem built in this way. It took us awhile, but some suggested Eclipse as a good example, OpenStack as a really bad one and Android as the killer one. I brought up my favorite, that being the Open-SaaS model in use by my good friend Raj @ Raintank and their Grafana open source ecosystem. But I’d say we’re on new ground in terms of betting on big valuation open source companies being able to monetize developer-driven IT stacks.
I think the way in which Docker, Inc moves into commercial objectives in the coming year will affect the outcome of technologies that are going to outlive the hype and be truly adopted in the business community. One can expect that the friction between Docker Swarm and Kubernetes, and perhaps Docker Engine and Rkt, will increase in the year to come as the business model tusses itself out.
In the end, I think my biggest takeaway from Barcelona was that very few companies have succeeded at building open-source companies on ecosystems as bold as Docker. And this, in my opinion, is what is causing the Docker ecosystem unease I noted above.
Is Docker going to justify its valuation based on on-premise enterprise IT installs of Docker Control Plane? Is it about owning the Registry and charging for images? In the builds (on prem or in the cloud)? Or maybe it’s in the SaaS-driven deployment? I’m not sure and, frankly, I’m not sure that Docker, Inc knows yet.
At the pace that this industry is moving, I’m sure it’ll get sorted out right quick. In the meantime, beef up on Kubernetes and let’s hope for collaboration, open-ness and flexibility for all parties involved. It’s a fantastic revolution we’re part of in the infrastructure world. Let’s keep it friendly!