The demise of docker and the rise of kubernetes


Docker was on every lips a couple years ago. Every small and medium company adopted it or was about to.

What’s left of Docker 3-5 years later? Well, not much.

Let’s rephrase the question to be more accurate. What company still cares about docker? Surprisingly the answer is nil.

The world has moved on.

Docker job prospects?

If you’re (only) a docker expert, you’re in troubles right now. There are no more jobs looking for docker expertise and you’re dangerously close to unemployable.

Docker expert meaning intimately familiar with containerization concepts and docker-specific implementations, as well as having experience building, running and debugging containers in production.

Some typical interview questions to expect (in 2017):

  • What is docker?
  • What is a container?
  • What’s the difference between a VM and a container?
  • What’s the good practice to build a docker images?
  • How to build a container?
  • Where are images stored?
  • What are cgroups?
  • Can you name some docker commands?

Here’s 6 typical questions being asked instead (in 2019):

  • What is a pod?
  • What is a deployment?
  • What is a stateful set?
  • How to update an application without downtime?
  • What is a namespace? When to use it?
  • Can you name some kubernetes commands?

The top questions are a piece of cake but all the Docker expertise in the world couldn’t help to answer the bottom ones. Those are kubernetes concepts, a whole new world of learning. Have a read, better keep up with current technologies to stay employable.

Companies are strongly biased toward kubernetes as of late. They’re really looking for the unicorn with kubernetes experience, preferably years, in production.

Kubernetes

Kubernetes has changed the landscape almost overnight. In the process, it also revamped every term around containerization and orchestration (see the questions above for some examples).

What was once called docker is now only referred to as a “container engine”. This neat marketing trick has been an outstanding success. Like an engine running a car, the container engine is still there, in fact it is running absolutely everywhere, including alongside kubernetes as a dependency, but its presence is ignored and has been forgotten by all.

Walk into a meeting, a job interview, a conference… kubernetes is omnipresent but the container engine never comes up anymore.

That’s the visible part of the iceberg right there. Kubernetes is integrating the whole ecosystem vertically and becoming the go-to solution.

Behind the scenes, it’s a subtle move repositioning Docker as a replaceable engine, waiting to be pushed out. There are multiple actors at play incentivized and actively working to take it out (RedHat, Google, Amazon, etc…).

Container management

Kubernetes has seduced management

Kubernetes has succeeded where docker failed. Management buy-in.

Working in and around F500. Pay attention to top management meetings, all-hands, announcements and other large corporate events. That’s pointing to where the time and the money is going to, paving the road to the future.

Well, half the time it’s paving the future, the other half is getting budget and paving self-promotion. Either way, that’s the right road.

Some of the strong recurring themes are #Cloud #Kubernetes #AWS #AI #MachineLearning #BigData #BlockChain. (Take a guess which one is a game changer and which one is a fad)

Counting how often kubernetes is mentioned, it is almost on every event, repeated many many times over. Whereas the container engine has been mentioned exactly zero times since the beginning of the year. (Still on the lookout for it, there’s 2 months left in the year.)

F50 are heavily invested or investing in kubernetes. More headcount, open hiring, consulting business, 6 and 7 figures checks going to enterprise providers and contracting agency. Not a penny of it is going to Docker.

It’s glaring obvious at this stage that docker has utterly failed to capitalize in the enterprise.

Where’s the money going?

First and foremost, “DevOps” people, whether that’s full time employees or contractors or consultants. Remember that human resources are always driving the project and the costs, unlike software license or hardware. Short parenthesis for workers, lots of opportunities in contracting / consulting. Obviously the market is especially hot for ex-googlers or other who contributed to these technologies.

For smaller and medium companies, usually on the cloud. The big money is flowing to AWS EKS, AWS ECS, Google GKE or Microsoft Azure.

For medium and large companies, on-premise or hybrid. The big money is flowing to VmWare (VM are still strong), Pivotal Cloud Foundry and RedHat OpenShift.

For enterprise consulting, the big dog seems to be Heptio, just bought by VMWare for $400M (a wise acquisition). On a side note, it shouldn’t take long before VmWare rolls a fully-managed kubernetes solution out-of-the-box on top of vCenter.

Kubernetes needs a container registry as well. Managed-solutions in the cloud have their own registry included, another line item in the billing.

On premise, either it’s the free registry, or a commercial solution when money is on the table, Quay seemed to be the dominant one for some time. The registry market might be a bit saturated lately though because everything jumped-on to be a registry including gitlab, github, nexus and artifactory.

The container engine registry is available in a commercial edition. Strangely enough never found it in any company or meet anybody using it. Really wondering if it has any customer or revenues. This might be one of the biggest commercial failure, right on the level of Apple map.

podman says hello

Since all the bits are available separately through different solutions, more or less managed. The only thing left to replace is the container engine itself.

Say no more. Challenge accepted! RedHat is killing it from RHEL 8 onward, transparently replacing it with its own engine. The name is podman.

Is it possible to kill a command? or more than that, a brand? or more than that, a vendor? Just like that?

Yes, it is. It’s been done multiple times in recent history. One of the most notable examples is mysql. MySql was acquired by Oracle somewhen around 2010 and subsequently dropped and killed by the community. The fork is called MariaDB. If you’ve run a “apt-get install mysql” in the past decade, high chances it setup MariaDB instead, getting aliased and substituted transparently.

Long story short. It’s a walk in the park for RedHat to do the same with “docker” => “podman“.

Who’s gonna buy Moby?

Totally $272M in funding over 9 rounds. The sell price would have to be quite high for investor to make a return.

That being said. Don’t fall into a common mistake of thinking that companies fail and become worthless. An unsuccessful not-growing company by Silicon Valley standards is still a very respectable medium business. A regular business could be worth $100M with a few hundreds employees. Talent acquisitions or killer acquisitions can do $1M per head.

Current headcount shows the company in the 100-250 range. No matter what metrics is considered, the company ain’t worth multiples of what it raised.

The worse may have yet to come though. The trend may be going downward rather than upward and I’m sorry for employees reading this. Employees are leaving, they can see the tide turning through the bad press and the better opportunities elsewhere.

The only thing of value might be the public image registry. Personal data and contacts from all members and organizations registered (sales lead or other monetization). Plus a direct automated deployment pipeline to almost every company in the world (the potential for dark patterns and malware is endless). Recall SourceForge sold for $20M in 2012 to distribute adware.

All things included, there’s some value to recoup. Just closer to 8 figures than to 10.

Call me Moby

Even if there was a buyer, assuming one of the usual suspects, RedHat/Pivotal/VMWare/AWS/Google/Microsoft. Why buy something today when it will go for half tomorrow? Why buy something when you already sell substitutes products/services better positioned? There is no pressure for anybody to acquire really.

Advertisements

15 thoughts on “The demise of docker and the rise of kubernetes

  1. I don’t like the tone of this article, but I think you’re mostly right. The industry (Google, etc) have invested (sorry, donated) a lot of money and resources to their CNCF foundation to make sure that Docker Inc looses its foot in the door to the Enterprise market. Why did they do it? Well, Docker is a product (install and run it, buy a license for more features, buy a support contract). Kubernetes is an open source project (not a product). Sure, you can install and maintain it yourself with kubeadm. But to really use it for stable systems you still better buy a product. And these players all offer such products based on Kubernetes (GCP/Azure Kubernetes, AKS etc). And because mutli-cloud is a thing that customers want there’s even space for new companies like Heptio. No surprise that they are founded by and operated by ex-employees of these large companies.

    Docker Inc. is doomed and the only concern of the industry is what will happen to the container registry if they go down. Maybe it will die and “pay for storage” registries (GCP, AWS, etc) will become the new standard. Or maybe it could get donated to the CNCF. I don’t think the container registry really is an asset for Docker Inc. It’s mostly a liability. It must cost a lot to operate it, and cost probably increases with increasing popularity Kubernetes and the cloud.

    So the big players in the cloud industry teamed up as a cartel to extinguish/kill/murder a startup with its “killer” tech. And they did it in a way that will never trigger antitrust because it’s all open source and anyone can participate (heck, even Docker Inc. is a lowest-tier member of the CNCF).

    The perfect murder!!

    Like

    • Thank you. Mostly agreed.
      Reworded a few things, you may have another read. Can’t believe this went live on HN 5 minutes after the first draft was published.

      I would add that Docker has missed quite a few steps, they were far from spectators to their fate.

      Like

  2. This article is a surprising page-turner! I am an editor of InfoQ China, I’m this close, this close to translating it into Chinese and reach our China readers! Without a doubt, the Chinese translated editon will add the URL and title of the original! If it’s ok, reply to me, thank you!

    Like

  3. I’m actually new to this world of tech. As such, this was eye opening for me. Any recommedations on where best to start my Kubernetes training?

    Like

  4. Cool post, you called that Kubernetes was the future so I guess that makes you Mystic-HFTGuy.

    You implied BlockChain may be a fad, any thoughts why this could be?

    Like

    • Blockchain itself isn’t a fad: it’s been the core of Git (and some other distributed VCSs) since before cryptocurrencies existed, and Git alone is far more widely used than all cryptocurrencies put together. (It’s an interesting exercise to look at the components of a cryptocurrency, such as blockchain, a distributed ledger, and consensus systems, and consider how use of a Git repo implements these.)

      There’s certainly a faddish bubble around cryptocurrencies, but the core technologies it uses and the way it uses them are used in other products and have been well accepted for years.

      Like

  5. Well, this not entirely unexpected if you look at what Docker is. Containers are just processes with special configuration; Docker does not run the “containers” themselves but merely configures the kernel to do so.

    Orchestration tools such as Kubernetes, too, are configuration tools, but in this case configuring Docker to configure the kernel. Once that layer is in front of Docker, Docker itself is heading toward becomming a commodity: any other container configuration tool that can do the same work can replace Docker, and the orchestration tool itself can even take over at least some of the configuration that Docker was doing.

    So Docker is being pressed on the one side by orchestration tools that also do configuration into the wall of the kernel actually running the containers, which Docker can’t do. This squeeze doesn’t leave it any room to grow, except to try to compete with the orchestration tools. Once Docker let those get ahead of what it provided, the writing was on the wall.

    Like

  6. Sorry but most people using Kubernetes uses Docker and there are lots of jobs that require Docker https://www.linkedin.com/jobs/docker-jobs/ as a DevOps engineer.

    So no Docker ain’t going anywhere, just because they can’t find a business model to fund themselves doesn’t mean they are going down the drain.

    This article sounds like it’s clickbait. 😉

    Like

    • Clickbait aside, Docker is already downsizing.

      With no revenues and unable to raise another hundred million as a shrinking business, it might actually go down the drain sooner rather than later.

      Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s