Can anyone tell me what the difference is between a "Residual" and a Join predicate.
If I haven't missed a subtlety they are exactly the same.
If so why confuse noobs such as myself by using two different words for the same thing.
+3
A:
These are distinct concepts. We're right into the details of of how SQL Server plans its queries, not the place for the faint hearted! But very interesting, you won't be a "noob" for long if you're studying this area.
This article describes how some predicates can be interpreted to "seek", first excluding many rows, so that only a few need to be examined to determine if they match the remainder of the predicate - the residual predicate.
This article describes how joins are processed, and how the details of the query can lead to the decision as to whether there's a join predicate, radically affecting the performance.
djna
2009-08-06 07:29:56
Can I just say you rock for that link.It makes sense too, I have an index ordered down date, which I'm doing a less than filter on. So the residual was using the search order to say, from every row past this point perform the join, thus saving computation by not having to perform the date comparison for every row.Awesome!
Spence
2009-08-06 07:48:32