Docker Cloud Pricing

I would like to make a few comments in response to @borja’s post:

1 - Docker Cloud is a service. My understanding is that Docker’s operation costs of running 100 containers on 2 large nodes is very similar to running the same containers on 20 smaller nodes. So why is the pricing different? Please correct me if there is a huge overhead when a user brings on another node.

“Over the past two years, as we’ve built this product we have queried and surveyed hundreds of …”

2 - What are these reasons? The Docker team have made little effort to communicate with the existing Tutum community during the launch of Docker Cloud this month.

3 - There are a number of existing solutions which have feature parity with Docker Cloud. These require a master/control node to orchestrate the agent nodes. The cost of running the master/control node (even in HA) is far less than $15/node/month.

4 - Making the argument that Docker has “great and free open source software” is both irrelevant as Docker Cloud is yet to fully integrate with these tools and a bit of a joke seeing as Docker Cloud is nearly 100% closed source.

Fundamentally, I agree w/ @latenssi. We are a fast growing startup that will soon be priced out of using Docker Cloud because the alternatives are far cheaper and have feature parity.

@revett

I’am senior developer at the second largest employment site in Australia (although we are still small), we are undertaking a rebuild of major parts of our API and mobile apps.

This afternoon I approached finance with a scoping proposal that would have integrated Docker Cloud into our workflow. I was denied, on one simple fact, it is far cheaper for us to develop our own solution on swarm and compose. Sure up-front we pay more, but we crunched the numbers and very quickly it would be more economic to have our team move forward without the massive cost overhead we would incur with DC.

Docker Cloud would be more then 3x our entire infrastructure cost. Read that again more then 3 x what we would pay for actual servers…

Yes, Im being strong, but i feel very jaded and dis-heartened by this move. Tutum was a product that had us excited, we have been let down. We contributed to Tutum and now our contributions have been abused.

I will be posting a Medium article with a breakdown of the cost analysis, relating the above: After less then 6 months it would be cheaper to roll our own in-house solution.

4 Likes

I feel being hurt when I saw the pricing, as a beta user, too.

maybe a good solution is: give the beta user free nodes quota, by 2 x (the largest number of nodes when we use tutum).

then I believe the old communitees will feel better. ( but I dont know how the new comers’ feeling… )

1 Like

To chime in again:

Our Infrastructure:

15 x $10 Digital Ocean Droplets (Singapore region) == $150
2 x t2.small AWS (Sydney region) == $58.56

Total == 208.56 per month

Our cost analysis of multi-cloud deployment pipeline with Docker Cloud:

Per Month - $255
Up Front - $0

Our cost analysis of a multi-cloud deployment pipeline with Rancher:

Per Month - $0
Up Front - $1,920
A one off cost for 16 hours of our lead ops guy setting up a Rancher + Drone CI/CD pipeline @ $120 per hour.

Payback period in comparison to Docker Cloud:

$1,920 / $255 == 7.5 months to payback.

Rancher also allows us FULL control of our pipeline, its open source and seems more compatible with compose at present… Basically with Docker Clouds pricing its a no brainer.

(PS Ranchers tooling, Interface and CLI are worlds ahead at the moment)

Seeing as @borja & @kickingthetv seem to be ignoring messages on the Tutum Slack.

I’ve been informed that Docker Cloud charges users for nodes which are "Unreachable".

Can you please confirm if this is the case and why!?

@revett

What I see these days is not only Docker jumping on the enterprise bandwagon, it seems to be a common new policy (IBM BlueMix did it, and others do it as well).

What imo gets forgotten is, that there are lots of indie developers out there trying to find an income using interesting and modern services (as Docker). Unfortunately they’re locked out by services/pricing targeting enterprises only.

To me it looks like this is a new and common strategy where quite some (former) lean startups (like Docker) head to. This movement makes it hard for indies to play with these now paid services and try to create something useful (instead of ‘just testing’ on a single node). This is a bad thing for enterprises as well, as it will get harder to find experienced devs and most likely there will be a point where another (open-source) service will launch and try to compete…

Maybe DockerCloud should think twice, because BEFORE they got funding, they were in the same situation as other indies.

But for now, the trend locking out small businesses from cool stuff continues…

I hope I could make my point clear (my english seems to get worse over time)…

2 Likes

The Founders of Tutum are in SF this week so most likely very busy, but I have been told that they will reply.

@revett

Thank you, we do appreciate the current situation and growth pattern of Docker is an immense undertaking. Although I will plead our case:

We are Ethical Jobs, a small non-profit organisation with the goal of connecting highly skilled workers to NGOs, non-profits and environmentally minded organisations. Currently we work with organisations such as Red Cross, Green Peace, United Nations, Amnesty International, Sea Shepard and others.

We are undertaking a complete rebuild.

Docker Cloud offers a service we would love to leverage. With the current pricing structure Docker Cloud has excluded organisations such as ourselves.

Its the model of pricing that I believe is in error, not the cost itself. With this model smaller organisations are penalized, there are other pricing structures that are far far fairer.

We understand that Docker needs to be profitable, although think on this: You were once small as we are. Your service enables small organisations to play with the big guys on a level playing field, just not priced as such…

For those that are interested:

Hi there,

I’m beginning to build a new product and and company around it, and I was super excited to utilize not only Docker, but also Tutum/Docker Cloud. This pricing announcement has completely knocked the wind out of those sails. At this point, as a small SaaS offering we are building, it’s completely unrealistic to use Docker Cloud as my Docker management platform.

My application is expected to only need 512MB - 1GB of RAM, but I don’t want to be running 5 applications on one server because of the potential CPU load. My plan was to get boxes that could handle 3 to 4 times the number of apps’ on the requirements so each client had burst space in the event their users all decided to do something at once. I also need to have a Mongo setup, and as it’s cheaper to roll my own than use Compose/mLab I was planning on using Docker to help manage that. So on that let’s do some math:

$17/month/server T2.medium x 5 = $85/month
(if I pay fully up front for the servers on AWS and that’s 2 application servers to start, and 2 dedicated DB servers)

$15/month/server after the first one = $60/month

Total: $145/month (converted to Canadian dollars currently that’s almost $200/month)

This is before I add on CI/CD services. At this point it’s cheaper for me to spin up another T2.medium and run Rancher and scale that as needed as it would start at $17/month vs. $15/month/server. The scaling needed on Rancher seems to be a lot less extreme in costs to adding servers to Docker Cloud.

There is also the fact that Docker Cloud doesn’t seem as fully fleshed out either. Rancher offers integrated handling of SSL certificates, and has a built in load balancer you can deploy.

At this time I can easily say that at this pricing and the “benefits” it is extremely overpriced to small businesses. When I told my boss about it at my current full time job he scoffed at the cost and said we would likely make our own platform as @andrewmclagan said. But hey, maybe us start-ups and small businesses aren’t the ones you are targeting, but I still find it hard to believe that paying for a few big servers could ever be easy for a small company to do. In my case I’d need to spin up a new server, pay to add it to Docker Cloud while potentially also paying for the one I’m replacing. Then transfer everything, likely incurring downtime for my clients, and then put everything back up in the hopes that nothing messed up.

I’d have to do that after adding 2 clients until I got to the point where I needed a second server because I couldn’t scale any farther. In this scenario though I’d be trying to fit as many clients on one server as possible before upgrading and potentially hurting their experience so as to not incur further bills to keep my company afloat. That doesn’t make sense to me as a business.

1 Like

So, Docker is all about containers… but when it comes to pricing, they charge per node?

Am I the only one who sees this as a contradiction?

2 Likes

I don’t mind paying the price, it’s well worth it. I run a small business and with Docker cloud I am able to manage the infrastructure myself rather than pay for IT managers and DB admins and network admins and god knows what other admin titles. I can do more with my little company than my competitors can with an IT budget 7-8 times what I am spending. We are now competing in larger arenas because we can scale up applications and services faster and cheaper than my competitors. Long story short. $15 per node , I’ll pay it.

The only frustration I did have was that it wasn’t apparent anywhere in the Docs that I could not scale up my nodes without first paying for the additional node. I was pulling my hair out trying to fire up another node and wondering what the heck I was doing wrong. Turns out I wasn’t doing anything wrong, my plan didn’t allow for it.

In conclusion, I like the service so far but I would rather have the billing dynamic based on the nodes I add. If I add one node bill me for one, if I add 7 that month bill me for 7 but not allowing me to add 7 when I need 7 unless I first increase the plan is kinda yuck.

3 Likes

Thanks pkzad for your feedback. The scenario you described should not happen. We are trying to reproduce. You should have been prompted for the billing information when you tried to scale up or add more nodes.

@ipedrazas Agreed, the concept of a container pricing schedule resonates better with me.

I have said it a thousand times: How can $15 be charged for say a $640 droplet and also for a $15 droplet?

I just disagree with charging upon infrastructure when Docker Cloud is trying to abstract that very infrastructure away.

Perhaps Docker could reap the same financials or even better from a container based pricing model?

I had setup my first test node using AWS. I setup a stack, tested it and
then terminated.

Using the web UI and following the documented instructions, I tried to
bring my own node. I copied the script that was generated, ran it on my
local host via CL and would get nothing.

I read all the the details about opening the ports so I logged into our
router and set the TCP/UDP ports as recommended. I then tried again from
the web UI with no result.

I read further about launching from the CL and dockercloud agent did as it
said and removed docker engine etc and installed docker cloud agent but
still nothing showed up on the web UI. At this point I still had the AWS
node and that was all.

I opened the log file on the host and saw that it tried registering
several times but would fail and state error ‘402’.

Finally I was reading through the forums and couldn’t find anything under
registration error 402 but I did find someone complaining about the price
when I searched just ‘402’.

That’s when I read that someone else had a similar problem and it was due
to their plan size. I figured I would try terminating the test node on
AWS. A minute or two after I terminated the test node and refreshed the
web UI, the local node showed up in all its glory.

That’s pretty much the steps I did to experience what I experienced. Hope
it helps.

Please let me know of there is something wrong with my account that I can
fix that limits the ability to scale nodes, that will really be easier if
it truly is possible.

Thank you kindly,
Patrick Casey

So what is the current situation from Dockers perspective regarding the actual topic of this post?

This is the most talked about topic in the forums and the overall consensus among users seems to be that the pricing model is weird. @dockersupport Still no fullfilling answer from Docker Staff.

1 Like

Small startup here, not live yet. Part of tutum beta program, lover of deploying lots of small, single purpose units on lots of small, distributed droplets (it’s very resilient). At this pricing structure, I am (sadly, but definitely) leaving Docker Cloud.

1 Like

Was told on February 23rd that:

“Fernando or Borja will be replying”

Nothing yet, so we’re trying out Rancher and will move if successful.

@revett

That is what I am gonna try out. Rancher looks promising.

1 Like

Hello everyone, Borja here. I’ve been wanting to chime in here for a few days now, and I apologize for not having had the opportunity to do so earlier. But first things first, I want to let everyone know that even if we’re not engaging here daily, we are indeed reading this thread, and your comments and concerns have been heard loud and clear – you can rest assured of that.

Nobody today would argue the positive effects Docker containers have had in our everyday lives dealing with software. Seldom does a technology have an impact on its industry like Docker has. And while disruption and innovation are always welcome and appreciated, they often come bundled with their own complications. One such complication is pricing, and how one goes about calculating the value derived from a new product by a diverse set of customers and use cases.

I want you all to know that we thought long and extensively about pricing before we launched Docker Cloud. Our current pricing model and future pricing plans (Docker Cloud is just getting started) are NOT the results of a quick brainstorm session to get the product shipped as soon as possible. On the contrary, this plan was put together after countless months of internal research, user interaction analysis, customer surveys, user interviews, external third-party market research, user testing and extensive customer validation. As a matter of fact, during this process over a dozen models were proposed, of which 5 distinct pricing models were considered, each with their own number of different tiers and permutations.

During this time, we came to realize that this indeed was a complex problem to solve, as there is no apparent winner nor fairest model to govern them all. In the process, we considered container-based models, as well as memory-based, CPU-based, subscription-based and multi-dimensional pricing models, among others, before we settled on a node-based model. They each had their own pros and cons. We were not trying to create an innovative pricing model and instead were focused on offering something that was both familiar and worked for a majority of our users. And to somebody else’s credit, it wasn’t us who first put a node-based pricing model into effect in this space.

With all this mind, I want to stress how much we can actually relate to many of the comments and arguments on this thread, emails and private messages you’ve sent our way. This is because each and every one of those scenarios and use cases you’ve brought up was considered during our planning. Ultimately, and very analytically, having reviewed hundreds upon hundreds of use cases, it was deemed that the node-based pricing model was best for our users. There were alternatives that could potentially have made more economical sense to Docker, Inc. that were discarded in favor of best satisfying the anticipated needs of our customers today. So, while some may argue that a container-based model is better fitted for the company that popularized the application container, those use cases spinning up hundreds of containers every few seconds across a handful of nodes would clearly disagree.

As a company, Docker aims to provide tooling and services to cater to and cover the end to end software delivery lifecycle, as manifested in the “Build, Ship, Run” part of our slogan. As per your feedback on this thread, we understand that our current Docker Cloud pricing isn’t accommodating to the demands of hobbyists and small projects. We value your business and want to see hobbyists and startups alike thrive using Docker Cloud. We’re continuing to explore viable options and tiers to ensure Docker Cloud caters to and supports as wide a number of use cases, team sizes and project sizes as possible. As our pricing plans evolve to accommodate for these additional scenarios, we will communicate any changes to our users through the normal channels (email and in-app notifications, blog, forums, etc.). I’d also like to take this opportunity to invite everyone to check out the free and awesome open source projects available today from Docker, such as the Docker Engine, Swarm, Compose and Machine. Docker and its great open source community have all put a lot of effort into these tools and we hope everyone can benefit from them.

In the interest of continuing to make the best possible use of this forum, I will be locking down this thread. We ask that you open a new thread per topic, question or concern that you may have that warrants a general discussion if one does not yet exist.

As always, thanks for your ongoing support.

2 Likes