Adam,

Thank you very much for this article. I have been searching around for a more intelligent way to handle massive, bulk-loaded datasets of user information. We have customers that basically send us entire refreshes of their data on daily and weekly frequencies and other that only send us updates. I need to speed up this process tremendously and stop performing delete then insert operations on entire groups of users. Instead, I need to post updates only when data really changes from load to load.

I was particularly thrilled with some of the speed enhancements suggested by others of using the CHECKSUM( ) over the range of data as opposed to string compares everywhere. I will be testing it out both ways and see if I can quantify the differences I see in performance. Some of our data loads are small 2,000 or less, but we have some problem-children in the 200,000+ range.

Very good article, and I will make a point to search for more of these from you...your coding style makes it very easy to read and understand the T-SQL code.

Larry