Comments on: Identifying Duplicates: Traditional vs. Analytical.

By: Alex Nuijten

Alex Nuijten — Sun, 07 May 2006 15:02:21 +0000

First of all, thank you for your comments.

To Mr. Ed:
My response to your comment is included in the blog. Because it is not possible to format the TKProf in this comment section (markup is not allowed) I added it to the blog.

To Karl R.:
The database I use on my laptop, where I did these tests is a 10.1.0.2.0 (10g Release 1). It’s the Personal Edition that I have.

By: Karl r.

Karl r. — Fri, 05 May 2006 15:23:00 +0000

Very nice article.
You see in RowSource that a window sort does not reduce the number rows as a group by does and that’s what i think the direct write read is probably the storing of the result of the window sort operation in temp.

Greetings
Karl
PS.: which database version you used?

By: Mr. Ed

Mr. Ed — Fri, 05 May 2006 09:01:33 +0000

Something is fishy about your tkprof. Look at this line:

229376 WINDOW SORT (cr=1383 pr=1809 pw=431 time=3154001 us)

The “time” is over 3 seconds. That’s longer than the elapsed time for the entire query!