Just came across this article - anyone have idea why docker would be struggling? [Docker, once worth over $1 billion, tells employees it's trying to raise cash amid 'significant challenges'](https://www.cnbc.com/amp/2019/09/27/docker-is-trying-to-raise-money-following-arrival-of-ceo-rob-bearden.html)
Docker containers are a nice to have, VM’s provide many of the same advantages and more. Persistent storage in Kubernetes has been problematic for years. Only the largest corporations can afford to containerize. Plenty of applications were written so badly that they would need a rewrite just to function in a Docker environment. Even though Docker has been around for several years now, professional developers are very slow to make use of it.
I dunno, those are some of the technical issues, no idea what the actual financial issues are. Maybe CNCF already soaked up the market.
Because who is gonna pay for docker? Not very much people
Who would pay money for docker? Not any serious enterprises.
Docker have (unfortunately) failed to really capitalize on the early hype. My guess would be that they’ve had offers but (like a lot of companies) they thought of themselves as more valuable than they were. Also they must have a truly horrendous cloud bill every month, given the scale of Docker Hub and the fact they give away build services for free.
With that said I’d be *extremely* surprised if one of the large tech companies doesn’t look to acquire them. Docker is a pretty fundamental technology for a lot of companies these days, and whilst Kubernetes can run on top of other container runtimes, I’ve yet to see a production cluster that doesn’t use Docker.
My guess would be Microsoft, as they’ve invested heavily in Docker for Windows and allowing someone like IBM/Oracle/Google to buy them could be very bad news.
Other likely buyer would be VMWare given the number of purchases in the container space they’ve been making.
Kubernetes, Openshift.
Docker has raised a lot of money. They used that to try to move from being a tool to create and start up a process within a container easily, to being an enterprise product that tried to compete in what enterprises want, because enterprises spend money on products that solve their problems.
The problem is that they spent a lot of money building a large, stratified, top-down organization where everything waited on decisions from the top. That meant that when they tried to compete with existing companies like mesos, aws, coreos, cloudfoundry, etc. they were left behind: constantly trying to appeal to what they were seeing their competitors do, and eventually it all got thoroughly stomped by kubernetes which now is free, has support from every corner, and isn’t hamstrung by a single vendor.
Docker the company doesn’t have a compelling community (afaik wasn’t really a priority from the top, so community development was left to volunteers largely, outside of trade shows), has a lot of ill will for not listening to users about UI, bugs, needs, etc. It has, in short, stagnated and it’s painfully obvious to anyone who wants to make money by investing in companies that the valuation that was placed on docker earlier on just hasn’t been realized by leadership in the company.
Raising venture capital is going to be very difficult. They are too old for that. A vc firm or a bank with a loan request is going to ask them tough questions. “Why isn’t the company able to break even or generate net income from operations?” “How would I get my money back if I invest or loan money?” For straight debt, how would docker collateralize the debt?
Simple, Docker EE provides next to no tangible benefit for the 90% of users.
What does docker EE do that docker CE cant?
I think windows server integration?
The other feature docker sells is their container repo…but most companies can easily run artifactory (which will handle their internal node,pip,mvn etc etc deps as well).
So whats the point of their repo?
I legit cant see a wide use need for docker EE
Not sure what the communication barrier is. Y’all acting like I said Docker was untenable with no benefit. I merely stated that while Docker has benefits, VM’s already provide many of the same semantic advantages. I use both.
If you’re deploying a stateless application to a GNU/Linux environment, then containers are fantastic. If you want Windows support, BSD, non Intel 86 derived architectures, or persistence, then VM’s are going to be more practical. While containers are more efficient, VM’s are overall more stable and flexible.
Specifically, if you want to port software to a wider range of platforms, then you’re going to have to use VM’s, either with cheap tricks like qemu in Docker, or Vagrant setups. You can’t build both GNU libc Go binaries and musl libc Go binaries with just one container, you need two, with different base images. For Rust, C, C++, D, and the like, you can do a couple easy targets in Docker, and have to reach for VM’s for the rest. How do you capture a core dump in minishift? How do you run musl Linux guests on BSD? For these examples, Docker would eventually get there in terms of support, but these edge cases are not solved for containers today.
Obviously not all is happy in VM land; Fragility, slowness, incomplete support for extending VM’s, are barriers to success there. But the open fires in Docker land have only started to wane quite recently. I’d be surprised if CTO’s were scared to death to trust in containers until a few months ago.
What makes both containers and VM’s great is the discipline of specifying the complete deployment environment, as opposed to winging it and doing it manually, which is slow and risky. The real value is the complete environment, whether its provisioned by Dockerfile, bash, Chef, or whathaveyou. This grants operators far more accuracy than they’ve enjoyed in the past. Docker is simply one of many vehicles for delivering well-specified applications.
If you already have everything running smoothly in VM’s, then Docker would be an upgrade path for sure, in terms of performance. But for many teams, the return on investment may not necessarily be worth it. There’s a reason a lot of crap still runs on COBOL on dedicated hardware, even though technically speaking it doesn’t make as much sense to update.
Just as Docker eclipsed VM’s, I would say even leaner solutions like serverless, CloudABI, and WASM will soon eclipse Docker. It’s been one hell of a journey with Swarm competing with Kubernetes. If Docker has already had its day in the sun, maybe *that’s* the reason the money is flowing elsewhere.
In the article, the CEO says they’re looking for money “to continue to execute on our strategy.” But, what exactly is their strategy these days? They bungled the ecosystem by buying up all the interesting projects and then letting them die. Swarm (SwarmKit) had a rocky start and then stopped progressing. And, I don’t see what Docker EE offers that’s enticing for customers against other Kubernetes offerings.
I love using Docker, but I don’t know what their business model even is, anymore.
I can guarantee that there are investors willing to fund Docker, and buyer lined up. The problem is likely a mix of following.
1. The newer investors want a big piece of the company and the current shareholders aren’t willing to dilute their stake that much. Its possible docker has way too much stock issued.
2. The newer investors want preferred stock\* that gets paid out ahead of prior investors with preferred stock. The existing investors with preferred stock aren’t willing to go for it.
3. The offers for the company are such that the general shareholders wouldn’t get much money, but there are more general shareholders.
4. The investors and E staff are still clinging to an old valuation, and won’t accept investments and offers at the current vaulation.
5. Investors who have seen the numbers have decided that the company is doomed, and they do better to buy want pieces they want in bankruptcy.
\*Prefered stock has a minimum value per share. If there is 50 million worth of preferred stock and you sell the company for 40 million the common stockholders get nothing, and the preferred sotckholders get all 40 million. Things can get complicated with some stock being more preferred than other preferred stock.
Its main product, the Docker engine and client, are open source (so no competitive secret sauce), and it’s enterprise offering (Docker Enterprise Edition) is very weak, mostly because they built it on top of Swarm instead of Kubernetes (which they adopted way too late) and Swarm scales very poorly. This armchair CEO says they should have built a kick-ass Kubernetes offering from the beginning and added Docker extensions to that, like Hashi is doing with Terraform Enterprise (e.g. Sentinel)
I don’t blame them, though, for trying to make Swarm THE answer to container orchestration, but k8s literally came from Borg which started this industry
They really can’t compete against AWS EKS or Google’s GKE since basically no set up is required for either product and you get k8s whereas Docker EE requires a lot of setup in comparison and all of its internal components (network overlay, role scheduler, logging) are not configurable (at least they weren’t when I used it last year)
People shit on RedHat for moving away from docker but it makes a bit more sense now.
There are companies like AWS which have continuously sucked the blood out of open source and many such companies in the name of cloud services. As much as I love them because of my career in it, I don’t like what they are doing to the entire ecosystem.
Docker just didnt have a differentiated enterprise product; EE was too expensive, too unstable, and didnt offer enough value to justify it. That coupled with what seemed like glacial development timelines (product capabilities announced then released 1 year later), was their downfall.
Never-mind the fact that Docker damaged their community by fundamentally stopping any discussions around CE and controlling all meetups and even DockerCon to be almost entirely about EE.
Finally, they locked away so many features inside EE and didnt not provide a cheap on-ramp for people to get a taste and then upgrade through the versions. Should have released a version with all features, but limited to say 6 hosts, or released a version that only included ucp, but no.