The nonclustered index is narrower, so fewer IOs if all that's needed is that single column. The clustered index has the entire data pages at the leaf level, so it's...

It seems that the file name you are looking for contains yesterday's date:

Today: 3/1/2012

Expected Expression Output: Filename 02-29-12.csv

If that's correct, I'd suggest first subtracting the day to get the...