Hacker News Reader

They are bloated and they are not performant. If you are using Node and the main benefit of node is its asynchronous execution flow you would be shooting yourself in the foot by using an ORM and gaining all the overhead that comes from serialization and deserialization.

View on HN

Your comment reeks of inexperience.

Haha, your comment reeks of inexperience. The thing you have namely not experienced is the pain of having to port an entire service to a different language because SQLAlchemy is as slow as a snail.

But yes, keep on using your ORMs because using a query builder is too hard. I'm sure your significant experience building services that will at most ever handle 2 QPS is very important to the software engineering world.

For what it's worth, if SQLAlchemy (or any other ORM) is slow, it's probably not because of "serialization" and "deserialization" (of whatever you had in mind), but because the database is queried in a (really, really) inefficient way.

It's also unclear how that's related to whether you're working in an async environment or not.

He was probably thinking about stuff like that.

One of the biggest bottlenecks for us was creating these massive nested Python objects.

Reply on news.ycombinator.com