For a data analytical web app, would it be better to use django's auto-generated relational objects or raw SQL queries?

Question

I am making an application that will serve as a demonstration of the data analysis capabilities our team can provide. I am very new to django and a the auto-generated APIs seem pretty cool, but I worry about scalability and having the quickness that only a carefully constructed and queried database can provide. Has anyone been in this situation and regretted/been satisfied with there choice of raw queries vs django APIs?

Answer 1

Django's ORM is great. It is one of the most complete and easy to use ORM I have ever encounter, but as every ORM, it has its limitations.

If your application requires full control of the database and very efficient queries, you might consider other approaches and compare them to Django and see which one fits better. You'll have to do some research.

Django is great for developing very fast complicated database applications, but I'm sure --if the application grows long enough-- sooner or later you'll have to start working directly with your database engine for optimization reasons. ORMs are generic tools, so database engine specific functions will not be available.

There is no rule to decide wheter or not django is gonna work for you but, one thing I can tell you is that its ORM helps you get your app started very quick and if you find some specific circumstance where you need to customize your SQL, then you can do it in Django as well. If not, just create a Python module which handle the database as you like in those specific circumstances and use it from your Django code. That probably will be the best way if you need to show your very efficient data analysis capabilities.

I hope this bring some light, it is a very wide question. One thing I'm sure is that you won't regret until your app grows big enough and when it does, you'll have the resources to find great programmers that could twist Python to handle every specific situation that behaves odd with Django's database accessing tools.

Found this link which may be helpful.

Good luck!

Answer 2

I'd consider this a case of premature optimization.

Django ORM is good enough in a general case, provided that your database is reasonably designed, has appropriate indexes, etc.

In > 90% cases this will be adequate, and often optimal.

When you will have identified specific complicated and slow queries, have reviewed the ORM-generated SQL and came up with a better query, then you may add a special case for it.

Maybe you will have more than one such special case. I still think that ORM will save you a lot of legwork in the 90% of database access cases where it is adequate.

Besides querying, an ORM allows you to describe the DB schema, its constraints, ways to recreate it and migrate it between versions, etc. Even if the ORM would not let you query the DB, these management capabilities would be enough reason to use an ORM.