Ask HN: Is anyone being able to use Django+MongoDB in production successfully?
4 comments
1. As you said yourself, django no-rel basically means you're stuck with a version of django that is 2 point releases old and will have to forgo all future releases of django for the foreseeable future.
2. Using mongoose, however, means you throw out the admin ui and any other apps that use django's orm.
At this point, I think it is very clear that you have 4 feasible options:
1. Use Django with mongoose.
2. Use a lighter framework like Flask, since you kind of lose most of django's benefits when you throw out the django orm.
3. Use Django with SQL (preferably postgres)
4. Use some other language (Ruby?)
I recommend that if you really want to use Mongo you should use a framework like Flask. Otherwise I recommend you use SQL and stick with vanilla Django. Worry about scaling when it becomes a problem. Scaling is a problem you want to have! When I was working at a startup I wish our main problem was scaling (versus finding product market fit). Don't prematurely optimize.
2. Using mongoose, however, means you throw out the admin ui and any other apps that use django's orm.
At this point, I think it is very clear that you have 4 feasible options:
1. Use Django with mongoose.
2. Use a lighter framework like Flask, since you kind of lose most of django's benefits when you throw out the django orm.
3. Use Django with SQL (preferably postgres)
4. Use some other language (Ruby?)
I recommend that if you really want to use Mongo you should use a framework like Flask. Otherwise I recommend you use SQL and stick with vanilla Django. Worry about scaling when it becomes a problem. Scaling is a problem you want to have! When I was working at a startup I wish our main problem was scaling (versus finding product market fit). Don't prematurely optimize.
Oops, I meant mongoengine, not mongoose. I've been coding in node.js of late.
We are also on nonrel.
It is a bad decision to move back to PostgreSQL, better invest that time on your product and make it better. You can always change later.
It is a bad decision to move back to PostgreSQL, better invest that time on your product and make it better. You can always change later.
Given that you're an early-stage startup, I'd suggest you go with a combination that allows you to develop your product quickly. If django-nonrel isn't seeing much development anymore, it would be safest to move away from it.
Quickest (and easiest) to MVP would probably be Postgres but what when we scale up? I know Instagram did it, but I also read that Systrom went mad making calls to the best minds in the Valley during their Android launch while he was at it!
Drop Django-nonrel. It has issues of it's own. You can stop using Django's ORM and use Mongoengine[1] ODM which is actively under development and awesome. You might have to make some extra effort, but it'll at least keep you away from problems of some stale 3rd party app.
[1]: http://docs.mongoengine.org/en/latest/django.html
[1]: http://docs.mongoengine.org/en/latest/django.html
We're definitely exploring mongoengine. The problem is that we've read that it doesn't support the Admin UI then. All these tradeoffs!
It seems the only two ways to use Django+MongoDB together are either using the django-nonrel fork of Django 1.3 which has its own problems (e.g. pretty much no more development) or to use MongoEngine (active development but no Admin UI). Both of these are causing us to sacrifice features of Django at the very least.
Are there any applications out there that have managed to use Django+MongoDB in production successfully and are those apps still running? How have you gone about it?
Do you think at an early stage its worth spending an extra 1-2weeks shifting back to PostgreSQL and not worry about scalability and multiple requests? I'm not the most talented DevOps guy, I would prefer a solution that would abstract all that away as much as possible, but then again, am ready to do what's best for the product.