It would help to know how you are testing your server's performance -- then we could perform the same tests on our systems and let you know if the results are the same as ours or different.
For example "when I copy a file of 1,000,000,000 bytes from an ibay mapped to the F: drive to my workstation's C: drive , it takes xx seconds; when I copy it back from C: to F: yy seconds".
(Note: the above is a guess at what you are trying to speed up, since I can't tell precisely from your post).