Whoa! Thank you very much for posting the solution and it's real nice that a person on that thread recommended creating some XML indexes, but in the name of Codd and all that is holy in the world of databases, let's back the database truck up a couple of miles.

Ok... let's ask a question here. How many people think that storing a CSV column of data in a database is a good idea? 😉 Even if you could index it, would it be a good idea?

I certainly hope no one thinks so. So why do people allow what turns out to be some highly denormalized data with HUGE and (many times) unnecessary delimiters that will screw up their online indexing because of the blob content to be stored in their databases???

Forget about the XML indexes. Normalize the data and store it in the database correctly!

{EDIT} Yeah, yeah... I know 2012 overcomes the blob-content problem with rebuilding indexes online. What's been the excuse since 2000 came out? That doesn't change the fact that XML is one of the most horribly bloated, pipe clogging, disk hogging, memory eating, denormalized methods of communication ever devised (Heh... man, it was hard to keep THAT clean:-D). 😛