mklink to network share or UNC path or Mapped Drive?

Are there performance or permission or other considerations to think about when using a mklink path to a network share versus just a straight UNC path (or mapped drive for that matter).

For example, can these three ways of accessing a network resource be considered functionally equivalent and roughly interchangeable?

mklink /d c:\shares\warehouse \\server1\warehouse
xcopy /s c:\shares\warehouse d:\temp\warehouse_copy

.

xcopy /s \\server1\warehouse d:\temp\warehouse_copy

.

net use X: \\server1\warehouse
xcopy /s X:\ d:\temp\warehouse_copy

Server is Windows 2003, clients are Win7 Pro. Network is mostly gigabit, though there are few 100mbit laggards here and there. I used a cmd shell in the example because it's easiest to explain, in practice the resource would be accessed by a variety of other methods also (Windows Explorer, Office "open" dialogs, system backup services, etc.)

I would strongly recommend NOT using symbolic links that have a remote target. My rationale being that a symbolic link makes an entry in the NTFS Master File Table, and although not substantiated, I reckon this could cause issues when performing low-level NTFS MFT operations (such as an offline CHKDSK).

As for performance, I can't see that there would be any difference at all. Both result in SMB traffic. The symbolic link route has to go via a redirection (handled by NTFS.SYS), but the "latency" here will be many thousands times smaller than any subsequent network delays...