I just returned from two days in Seattle at DockerCon 2016. What I learned at DockerCon this year can be summed up in four categories:
- Container adoption is on the rise
- Docker is winning by making containers simpler
- Docker is forging a path to win enterprise workloads
- The battle for orchestration just became more interesting
Container Adoption Is on the Rise
We all already knew this, but DockerCon confirmed it. Just the size of the conference told the story of Docker’s rise over the last three years. The first DockerCon had 500 people. Last year’s show had roughly 2,000. This year there were 4,000 people in attendance.
Even more compelling is the growth in the ecosystem. Last year, there was probably about twenty vendor booths, and most of those vendors were early stage startups. This year there were rows and rows of vendors, including giants like IBM, Microsoft, Cisco, EMC, Oracle, HPE, and many others. Many of last year’s container startups are delivering robust solutions now, and a whole new breed of startups has emerged.
In previous DockerCons, much of the audience had deep knowledge of containers, because it consisted mostly of early movers. This year, we saw a lot of people from enterprises, and many of them were in the early stages of discovery. That is a good sign for the container industry.
Even more impressive are the numbers Docker shared during the day one keynote. Take a look at some of these staggering stats:
- 2900+ contributors
- 4.1B image pulls—up from 1.5B from January
- 31K Docker pulls—720% growth
- 95K projects on GitHub with “Docker” in the title
- 460K Dockerized applications—3,100% growth
- 250+ Meetup groups
- 125K Meetup members
These are astounding numbers considering the Docker project has been around for only three years.
Docker Is Winning by Making Containers Simpler
What is really amazing about the Docker rise to fame is that it did not really invent anything new. Containers have been around for a long time. What Docker has done is made it easy for developers and operators to run and manage containers.
This year’s conference showed us that Docker has made heavy investments to make container usage even simpler and less complex. As CTO Solomon Hykes explained in his keynote, “Before, you had to be an expert to run containers—now, you don’t.” Hykes talked about the options companies had previously. Option one was to hire a ton of experts in order to take advantage of containers. Option two was to hire an outside consulting firm, which locked you in to that vendor.
Docker addresses this issue by focusing on the user experience first and foremost. In fact, it hired UI experts from the gaming industry to improve the overall user experience. And when I say “user experience,” I am not just talking about fancy user interfaces. Docker has a very user-friendly API experience that hides a lot of the underlying configuration and integration tasks that formerly required an army of experts to figure out.
Here are some of the announcements that fall into the category of simplicity and usability:
As Hykes said in his keynote, “making things simple is very hard work.” Docker has put a lot of hard work in since the last conference, and the result is that working with Docker containers is much simpler than ever before.
Docker Is Forging a Path to Win Enterprise Workloads
Last year, Docker focused a lot on features to convince customers that containers were ready for production workloads. This year, Docker is focusing on convincing enterprises that containers are ready for enterprise production workloads.
To make the case that Docker is ready for the enterprise, Docker marched out enterprise customers GE, ADP, Kroger, and Health Direct Australia. The most compelling client case study was by far ADP. ADP processes over $1.8T in financial transactions and stores over 55M Social Security numbers. Very few companies are held to stricter security and compliance controls. This is a tremendous use case for both secure containers and containers at scale.
The release of Docker Engine 1.12 added a much-needed boost to the overall security architecture of the Docker platform. The security-related features in Docker Engine 1.12 include:
- End-to- end encryption available out of the box,
- Out of the box TLS configuration ensures all nodes in a Swarm communicate with each other using mutual TLS.
- Cryptographic Node Identity is a general-purpose framework that allows for trusted workload dispatch. It enables cryptographically secure decisions about which nodes can run or sensitive workloads or access sensitive networks.
- Seamless PKI with automatic certificate rotation.
- Configurable acceptance policies to configure how nodes can join a Swarm: automatic, manual, or require a secret.
The Battle for Orchestration Just Became More Interesting
Up to now, Kubernetes is by far the most advanced orchestration tool for containers in my opinion. Mesos usually ranks second and Docker Swarm a distance third due to its lack of maturity and feature. Docker Engine 1.12 beefs up Dockers orchestration capabilities. In fact, what was described in the keynote sounded very similiar to the Kubernetes feature set. Keep in mind that Docker’s orchestration is in beta mode, while Kubernetes and Mesos have been battle tested in the real world for a while now.
Here is a short list of new orchestration features:
- Self-organizing and self-healing capabilities
- Built-in service discovery
- Built-in, strongly consistent distributed store
- Consistency and resiliency of app in case of node failure
- Dynamic role promotion and demotion of Engines in a Swarm
- Zero downtime node management with maintenance mode and advanced diagnostics
Here is my take on orchestration: I work with several Fortune 500 companies. Very few of them are running containers in production, and even fewer are running containers in production at scale. What this means is that many enterprises are not using orchestration tools yet.
For those who do need an orchestration tool now, Kubernetes is the clear choice. However, I think that many enterprises are still twelve to eighteen months away from needing to make that decision. If Docker’s orchestration solution is ready for prime time by then, many enterprises may default to Docker’s orchestration tools because they are already heavy users of Docker containers. Docker also does a great job of making container technology easy to use. Unless Google focuses on making the tools much more user-friendly, there is a good chance that developers and operators will migrate toward Docker for orchestration.
I’ll be keeping a close eye on this over the next year or two.
Container adoption is on the rise, and Docker continues to dominate in this space. Docker excels at making tools very simple and getting out of the way of engineers so they can do their job. The Docker ecosystem continues to expand, and Docker continues to release features at a rapid pace, sometimes in direct conflict with its ecosystem partners.
I expect that in 2017, enterprises will start moving to container-based architectures in production in mass. I believe that for the rest of this year, we will see enterprises kicking the tires on Docker and providing feedback on the remaining gaps in the security that heavily regulated industries require. Docker will quickly iterate and release these features, leading to a surge in enterprise adoption in 2017.
DockerCon 2017 will likely be an enterprise production workload coming-out party. You won’t want to miss the customer use cases next year.