Ideas and Validation October 18, 2020

MongoDB as a service?

Shakil Ahmed @daemonunderbed

Developers, would you like a MongoDB as a service?

Services like vercel and netlify have solved the issue of easy deployment of APIs and front-end code, but most of them don't provide any database or the DB support is really subpar.

How about MongoDB as a service with a robust web-based GUI and $10> starter plan?

  1. 4

    Operating a database as a service is a race to the bottom. You'd be competing with Mongo's first party services, for one. And also AWS Document DB. You'll never reasonably make any decent profit selling the database hosting since your biggest competition is literally the manufacturer. And as others have said, there's licensing concerns.

    Plus, you don't want to be in the business of storing other folks' content. You inherit all the burden of security and compliance.

    In this thread, you mentioned other problems. A gui, autogen, etc. Why not double down on those? The margin is significantly better for you, and you can build great tools to spin up Document DB on AWS with Cloud Formation or something like that. "Click a button and you get a better managed database"

    1. 1

      @mattbasta thank you for your feedback. I will surely look into it!

  2. 2

    MongoDB has some limitations built into their licenses that you should be aware of: https://techcrunch.com/2018/10/16/mongodb-switches-up-its-open-source-license/.

    Not sure how some other providers are operating, tbh, with those restrictions.

    However, as an alternative perspective, I think you’ll want to consider monetization here pretty carefully. The real money for a DBaaS is in charging companies $1000s/month to run a scalable, secure, high-availability service. Making a downmarket play might work, but how many folks are there that think $60/mo for Atlas’s lowest production tier is too expensive? And, they probably only think that until their app is making money and then you don’t care about $60—you just need the service to be reliable.

    So, i think you should consider carefully if you want to try to compete on price as your USP. It might still be doable, but suggest modeling out how you’ll position the service relative to Atlas (the big player) and some of the other small ones.

    1. 1

      Spot on. I've been looking into similar opportunities a while ago and ended up with similar conclusions. Also, offering managed DBs on major public cloud providers will is a relatively low margin market while still competing with the cloud providers themselves.

      Btw, a rapidly fast-growing company that focuses on managed DBs and Queues is https://aiven.io/.

  3. 2

    There are developers (+168) who likes this idea, https://vercel.com/integrations/mongodb-atlas

    As a poor dev, I used to use free tier mLab with Heroku, recently acquired by Atlas. Atlas also provides a single cluster for development purposes. If I ever go to production I'd assume I'd spin up a DO node and install it over there. Or change my way to Amazon Services.

    But my use case a bit different. I'm not using it with full features just for convenience on json based document storage, that's it. I don't need to write models for the data no need to create classes etc. Take the data serialize and put it into the db.

    GUI is kinda overkill for me since there is apps for it to use locally, however it would be a nice touch in the future not necessarily a killer feature.

    I heard that it's $60/month, so if you can compete against. It may work and might be the next acquisition.

    1. 1

      I have to do some research before doing anything serious about it considering I didn't come across this solution before https://vercel.com/integrations/mongodb-atlas

    1. 1

      The pricing does not seem suitable for indie devs to me. Also, lacking in dev supports. For example, it could use a better GUI for viewing data. Some features like automatically generate controllers (express, Django Rest Framework) based on collections.

      1. 2

        It starts free. So you have a collection already and you want to generate a controller? how do you do that now?

        1. 1

          The free plan is not good for production, I would not build any serious app on top of the free plan.

          Right now, when I have a collection, I write the controllers for it. Frameworks like DRF make it really easy, but if it's generated automatically it's better (I think).

          Their shared cluster is $15/GB. If you could get it for $10/GB from a lesser-known provider (A startup), would you go for it?

          1. 1

            The difference between $10 and $15 is not of consequence to me if I'm making something for production and for customers. If I can't make $60+ within a year, the product is shit.

            1. 1

              I agree with you, I would not go for a slightly cheaper one either.

          2. 1

            It's very easy to add MongoDB as part of the infrastructure. Most developers don't find this a pain. The only pain is availability which is what users are paying for (Cloud Atlas).

            In terms of GUI, there's less value in that as most of the framework that's out there provide ORM for MongoDB.

          3. 1

            I would! but I also got $1000 in atlas credits from their student pack so I’m not worried about their cost for now. I think you might be onto something though, their basic m10 cluster feels a bit expensive (although I don’t know nearly enough about dbs to know this well)

  4. 1

    It's good idea but there are few in the market. So, if you can come up with something unique, it may work.

    These are existing players-
    https://geekflare.com/mongodb-hosting-platform/

Recommended Posts