Python ORM for massive data set

I have a massive data set of customer infor开发者_如何学Gomation (100s of millions of records, 50+ tables).

I am writing a python (twisted) app that I would like to interact with the dataset, performing table manipulation. What I really need is an abstraction of 'table', so I can add/remove/alter columns/tables without having to resort to only creating SQL.

Is there an ORM that will not add significant overhead to my application, considering the size of the dataset?

See

• What are some good Python ORM solutions?

They seem to be in favor of SQLAlchemy

There's twistar, a native ORM for twisted. It works on top of twisted.enterprise.adbapi.

http://findingscience.com/twistar/

I thought that ORM solutions had to do with DQL (Data Query Language), not DDL (Data Definition Language). You don't use ORM to add, alter, or remove columns at runtime. You'd have to be able to add, alter, or remove object attributes and their types at the same time.

ORM is about dynamically generating SQL and developer's lift, not what you're alluding to.

If I'd have to pick the Python ORM, I would pick SQLAlchemy.

But with Twisted you have to be cautious since Twisted is asynchronous.

See: Twisted + SQLAlchemy and the best way to do it