True, but not always true 😀

Before windowing functions came through in SQL Server 2012, some problems could only be done using two ways:

1. A lot of joining

2. Using a cursor

Number 1 is set-based, but scales quadratically in some cases. In other words, great for smaller sets, not so great for large sets.

Number 2 is, well, a cursor, but it scales linearly. In other words, after a while it becomes better than the set-based approach. Until windowing functions showed up of course 😀

There are some great examples in Itzik Ben-gans book about windowing functions.