>I'm curious, What is your reason not useing a view of the production table and skipping the second step?

My reason at the moment is that I only solved this last night (after about two weeks of testing different approaches and very late nights). My reason for possibly keeping this option is that with a remote stored procedure, I may be able to 'fire and forget' to finish the processing off. This may not be so.

Also, as I may discover tonight, it may be worth dropping the clustered index, adding to the production table, and then rebuilding the index. Another point is that the current table is two and a half million rows. We intend to grow by 300 percent in the next two years - this will make future updates add to a table of about eight million rows. If I can get SQL Server to do most of the work it should be faster!

I'm doing this at home so I can check it in the morning before I leave for work and try different things at my leisure!