yeah weve thought about that.

1) all of our Reporting sprocs have isolation level set to read uncommitted so that they dont block. We are no worried about irrepeatable reads, etc.

2) worst case scenario, a user that accidentally got navigated to server that had the rug pulled out from under them, they would get an error and when they retry, they'd be directed to the other reporting db.

3) none of our reports take more then a minute or so to execute. that was a major goal of ours with the reporting system in the first place. Our reports originally were timing out in production all over the place even after 5-10 minutes.....

Great questions

GAJ