First question, did the execution time change with the removal of DISTINCT with the original indexing scheme?

Next question, did either of those indexes get used? Did you check the execution plan to see if they were picked up by the optimizer or did it stick with the original index. If it didn't choose those, we might want to try an index hint here. But, this part, it's desperation to squeeze a little bit of performance out. It won't make a major difference. The real problem is the volume of data we're attempting to move. There's no way, within a query or index, to simply make that "fast." Moving that much data is largely dependent on hardware.