Why CoreOS just fired a Rocket at Docker

8 Comments

For the past several months, it seemed like Docker was on its way to becoming the de-facto standard for container technology, the hottest thing in cloud computing in 2014. Then along came CoreOS, which dropped a bomb (or in this case, a rocket) on Monday, kicking off what could become a container-standardization war between the two entities.

CoreOS’s announcement that it built a container engine that can potentially compete with Docker’s container technology caused quite a commotion within the tech community on Monday. Docker has enjoyed a swift ride to prominence this past year with its container skills catching on with some of the biggest names in the industry, such as Google, Amazon and Microsoft. CoreOS’s decision to detail its own container plans in light of Docker’s momentum is an aggressive move, one that highlights CoreOS’s belief that it can capitalize on Docker’s perceived weaknesses.

Alex Polvi, CEO, CoreOS

Alex Polvi, CEO, CoreOS

“We want Docker to be a component just like we want CoreOS to be a component,” he said.

Docker losing its way?

Docker’s effort to develop its enterprise-oriented lineup of features has caused the company to lose sight of its goal of making sure that its core container technology lives up to its promise, Polvi argued. As Docker’s recent acquisition of Orchard Laboratories and its Fig container orchestration service shows, Docker is working on technology to manage all those spun-up Docker containers, and it wants to make a business around that technology.

In this sense, Docker seems to be competing with container-management-and-orchestration services like [company]Google[/company]’s Kubernetes or the new [company]Amazon[/company] EC2 Container service, Polvi said.

“What happens when [the] Amazon container service is launching a container that has a distributed clustering system in it?” asked Polvi. “There’s tension. That’s why we built Rocket.”

The launch of Rocket comes at an interesting time for Docker, which is hosting its DockerCon Europe conference this week. Docker CEO Ben Golub even questioned the timing of CoreOS’s announcement in a blog post responding to CoreOS.

“While we disagree with some of the arguments and questionable rhetoric and timing of the Rocket announcement, we hope that we can all continue to be guided by what is best for users and developers,” Golub wrote. Docker declined to make Golub or other executives available for an interview on Monday, although Docker’s CTO and co-founder Solomon Hykes was noticeably upset about the way CoreOS unveiled Rocket and voiced his concerns over Twitter and in the comments section of Hacker News.

Docker Founder and CTO Solomon Hykes at DockerCon 2014

Docker Founder and CTO Solomon Hykes at DockerCon 2014

“‘Disappointed’ doesn’t even begin to describe how I feel about the behavior and language in this post and in the accompanying press campaign,” Hykes wrote on Hacker News. “If you’re going to compete, just compete! Slinging mud accomplishes nothing and will backfire in the end.”

It should be noted that CoreOS co-founder and CTO Brandon Philips has a seat on the Docker governance board, and that CoreOS “will continue to contribute and have folks on the governance board as long as we can,” said Polvi.

Hoping to contain it

With a new potential container standard hitting the industry, the most interesting cloud computing technology is about to get more interesting.

Pivotal released a blog post on Monday that explained how it was interested in collaborating with CoreOS “on an open specification for Linux containers.” The company appears to be hedging its bets so that it’s neither a full-fledged Docker or CoreOS supporter; it apparently just supports the idea of a container standard, regardless of which company is behind it.

From the Pivotal blog post:
[blockquote person=”Pivotal” attribution=”Pivotal”]Now that the industry is aware of and interested in Linux containers, Pivotal calls on the industry to create the POSIX equivalent for standardized containers that everyone can utilize with certain expectations and guarantees. Decoupling an open specification from the implementations decentralizes design decisions and allows the industry to explore the tradeoffs of those choices in more rapid innovation cycles.[/blockquote]

James Watters, Pivotal’s vice president of product, marketing, and ecosystem for Cloud Foundry, tweeted about CoreOS’s news as well and commented on Polvi’s claim on Docker removing its original “standard container” manifesto; this caused Hykes to strike back on what he said was not true.

For Oliver Thylmann, the founder of a container-management startup called Giant Swarm that’s similar to Kubernetes, the news of a new container standard is fascinating to observe, and his team is already looking at Rocket. Thylmann said that the Docker daemon — which is basically a container-management program that runs in the background — is getting bigger each day as the company adds more features, and that bulk is not something Giant Swarm necessarily wants.

“We don’t really care if [the containers] are Rocket or Docker or such,” Thylmann said. “It needs to be lightweight and stable.”

Regarding Docker pushing itself into being more of a Kubernetes-like system, which would be similar to his own startup, of course, the company could risk alienating its users who just want it to concentrate on the containers themselves, he said.

“I feel that Docker is creating enemies based on possibly doing too much and not doing the parts it is already doing not good enough, and that creates enemies.” said Thylmann.

Still, it’s going to be tough for CoreOS to gain the type of momentum Docker has going for itself. Docker has a large community and now boasts “over 18,000 project in GitHub with Docker in the title,” wrote Golub in his blog post. CoreOS has a long way to go before it can gather that developer support, although if it is indeed a more lightweight and secure container technology than Docker, it could get some traction.

At least one user of both technologies, who declined to be named in the interest of hedging their own bets, said that Rocket’s ability to “sign and encrypt images” is appealing and it seems as if CoreOS is attempting to make Rocket a more simple version of Docker.

It’s still far too early to tell whether Rocket’s container technology will catch fire the way Docker’s did. Regardless, the pressure is on Docker to distinguish itself now that it has a competitor that seems to be coming out guns blazing.

Barb Darrow contributed to this report.

8 Comments

TechYogJosh

This is a classical challenge in technology that a start-up begins as “changing” things making them simple, but then realize to make money, it need to make things heavy and perhaps more complex. To be successful commercially it need to address as many users as possible and that in many cases, much as alleged against Docker, go against the founding principle. However, a technology company is not operational to be a start-up always. Scale will come when the company targets all the potential customers. With this it may have to change its initial hypothesis. The industry need to understand and appreciate this than live in a romantic ideal utopia.

Mike

I have nothing against a lightweight, simple container, but let’s drop the POSIX-type standardization. Management standards are notoriously slow, lag the innovation cycle, and still aren’t standard. Let the management guys (Kubernetes and Giant Swarm) abstract the differences between containers.

Furthermore, I don’t see any problem with Docker adding features to be more scalable and manageable. If they want to become a Kubernetes, then why not?

In the end, let the market decide what they want.

AL

I think just upgrading the POSIX standard of LXCs to something more userfriendly like docker or rocket is the right way to go. Five years from now there should be only one container format that can run on Mac, Ubuntu or redhat. Same should be true about virtual machines.

Mark

I still am trying to see the value of Docker vs what application developers need – a PaaS. Containers are great for replacing VMs. We don’t need a replacement. We need to not care.

Richard Raseley

That is kind of the point – Docker is positioning itself (or trying to position itself) as the defacto standard upon which infrastructure-agnostic PaaSes are built.

Comments are closed.