How Docker is getting back to basics
The company that popularized cloud containers failed its enterprise play. Now it's turning back to the developers it began to overlook.
When the definitive account of cloud computing is published, Docker probably deserves its own chapter. Within the short history of this company you'll find both an enviable technology story and a painful business story. It's the job of Scott Johnston, its latest chief executive, to add a third verse.
Docker didn't invent containers, the novel way to package software applications that improves the speed and flexibility with which that software can be deployed. But Docker did come up with a user-friendly way to incorporate them into application development. Its container became a de facto standard as cloud computing hit an inflection point around the middle of the last decade, turning heads in tech and business as one of the most quickly adopted enterprise technologies in recent history.
Investors poured more than $300 million into the company, yet Docker never found a way to translate its technological success into a major business. It swung for the fences with an ambitious plan to become an enterprise software and services vendor focused on Fortune 500 companies, but for a variety of reasons, companies like Red Hat and Pivotal made more money selling container services to big businesses than the company that made it all possible.
Last November, Docker hit the reset button. It sold its enterprise business to Mirantis for an undisclosed amount, and named Johnston, who held several key product and operational roles over the years, as its fourth CEO in less than three years.
Now Docker is trying to reinvent itself by once again focusing on helping developers use containers in their applications. At a revamped virtual developer event Thursday, it is announcing a new partnership that allows Microsoft customers to access Azure resources from within Docker Desktop, its flagship user-friendly application for creating containers, and tightens the links between Microsoft's popular Visual Studio Code developer tool and Docker.
Ahead of Wednesday's event, I chatted with Johnston about the pandemic, Docker's missteps and what comes next.
This interview has been edited for clarity and length.
How has the pandemic affected the company?
We were about 40% remote before the crisis hit. And so we already had some muscle memory and some processes and protocols and communication standards that facilitated, call it a hybrid state. So for those employees that were 100% in the office, it was a little bit of an adjustment, but I'd say for the vast majority of the company, we were able to slip into full remote pretty easily.
From a business standpoint, it's been really interesting to see the evolution. Our price point on average on a monthly subscription basis is relatively low, it's 20, 25 bucks a month per account. There are much, much higher ticket items ahead of us that get scrutiny before they get down to us.
What is interesting now is this almost-overnight shift from offline to online activity, a huge shift of human hours and attention and dollars shifting that is driving this huge upsurge in demand for developing new applications. From that standpoint, it's been one of those almost surprising tailwinds: The whole industry is now pushing us as a society to get online, and new applications are needed to do so.
You and I have talked many times over the last few years and there have been many, many iterations of what Docker has been hoping to be. What happened? How did this company get to this point?
We could probably talk for hours and [over] several bottles of scotch on that one.
So, roll it back to February 2014 when I joined and there's about 20 of us. It was just myself, Ben Golub, one other business guy and then, you know, 16, 17 open-source engineers including Solomon [Hykes], our founder. That very focused group was very much focused on developer and government teams.
In some sense, this is a little bit back to the future. We're going back to the original love of the product and the original personas that love the product, which is developers, dev teams and that workflow that they just got a lot of value from.
We could probably talk for hours and [over] several bottles of scotch on that one.
I'm not dodging my own role in how events played out, but with the enthusiasm around the Docker ecosystem and with the customers in the headlines, we did bite off a bit more than we could chew, and we started chasing multiple markets. In addition to this developer business that we started in 2014, we then started getting into the enterprise business.
And when markets are growing that quickly, and you're a startup growing with those markets, it's very difficult in the fog of a startup to execute multiple channels, multiple products, strategies and multiple open-source strategies effectively, right? Because you just peanut butter yourself over too many threads and too many different priorities.
There's a long-winded way of saying that we had a good enterprise business, but standing back, was it taking advantage of this bottoms-up developer love that I've been referencing? Was it really connecting the enterprise monetization with that developer love, and then was it as differentiated as other go-to-markets such as Red Hats and Pivotals and others? That became a very expensive business.
I'm not pointing fingers, Tom, because I was in the seat at the time. The other thing I'll say is markets change extremely quickly. The Docker engine, while a completely new piece of technology in 2013, 2014, as it was open source and it was picked up by many, many an ecosystem, it quickly got commoditized.
Which is a good thing for the industry, right? Because it moved the abstraction level up. If you start talking about orchestration and we started talking about Swarm, it's like the level of problems started moving up the stack. But our monetization model from the enterprise business was tied to the Docker engine itself.
This is a kind of context for the conversations that we had six months prior to November, where we stood back and said, look, where's the real differentiated business and where's the differentiated value that is very hard for others to replicate your work, to actually build an exciting high-growth business around. And that led us back to the developer business.
There's another thread of discussion out there about the perils of VC funding and the expectations that come along with raising money. I think a lot of people feel that Docker was forced into a position where it had to scramble a little bit more for revenue earlier than it was ready because of the level of funding raised and the expectations around that.
Any time you take investors' dollars, every time you partner with investors, when they invest in your company, you're aligned because you're all shareholders. And so of course everyone wants to have aggressive top-line growth. Do you pursue it [as a] bottoms-up, developer-centric business or top-down like an enterprise business or channel does?
One of the reasons we made the decision we did to restructure in November was to give the company the opportunity to refocus back on developers and dev teams, because that is where we know there's opportunity.
Over the next couple of years, it's your job to think about ways to grow, right? What else are you thinking about in terms of potential areas that Docker could continue to make an impact?
You can look at the types of applications that we enable with Docker Hub and Docker Desktop. There are a lot of users out there who are already using us for big data workloads, data analytic workloads, and that's an area, honestly, that we just right now don't have the scope and scale to focus on. But that universe is just exploding.
In terms of providing tools to data scientists, they don't want to be distracted by plumbing and complexity. They just want tools that help them ask questions of the data and get the results back. Now that's also a crowded space, and so there's other stuff in there that we'd have to be thoughtful about in terms of where our value add is.
Never in our careers have we ever seen [new] development primitives stop, right? There's always a new stack evolving, and there's a bunch of stacks that have all turned out to be blind alleys, you have to be thoughtful about sorting through all that.
But we see our role as coming back to kind of home base. We started our role as really making these development teams incredibly productive at building great apps. And so if it's containers today, what does it get augmented with tomorrow? What does it get augmented [with] the day after that? Keeping up with the different primitives that are relevant for developers and making it easy to consume, it's really important.
We've also talked about development teams, that's our unit that we obsess around in terms of features, functionality, ecosystem partners that augment the different stages of the tool chain. But imagine that we're now able to, because of the SaaS product, to see that we have 20 of these little teams landed at Nordstrom's, I'm making this up.
Then we come in a couple of years from now and say, "Hey Nordstrom CIO, you've got 20 of these teams. They're all running on credit cards. You've got no compliance, no visibility, no auditing. How about if we rolled all this up for you, gave you a single billing statement, helped you with the security."
This is a play that's been run by GitHub, Atlassian and Twilio. You let the self service developer pole land and basically seed these enterprises. And then there's phase two, where you come in and offer additional value on top, but it's really speaking then to the centralized enterprise organization.
All that functionality is a lot, and it'll take us years to build. But you'll increasingly see this additional line of revenue that's going to be much more about talking to the centralized IT organization, and rolling these all up into a consolidated enterprise sale.
Correction: An earlier version of this article misstated the day on which Docker's virtual developer event is being held. It is Thursday, not Wednesday. Corrected May 27, 2020.