This set based method is using an inefficient triangular-join. There are other set-based methods that will blow this (and, obviously, the cursor) away.
See http://www.sqlservercentral.com/Forums/FindPost728664.aspx for a set-based example of a one million row running update that runs in 35 seconds, and a corresponding cursor-based solution that runs in 144 seconds.

[Reply:
Hi wgshef,

Thanks for your comment. The technique that you refer to as a more efficient set-based solution is one that I wouldn’t recommend using. I don’t have enough room to elaborate here; please see my response in the following blog entry: http://www.sqlmag.com/Article/ArticleID/102251/sql_server_102251.html.

Cheers,
BG]

WAYNE6/4/2009 4:16:27 PM