sas/sata topology vs performance?

[Nicholas Leippe]

On Thursday 25 October 2007, Steven Alligood wrote:
> Without having any more information than this email (was there an
> earlier one I missed?), I would grasp at the straw that the different
> sides of the array are going through different controllers (either
> within the array or in the attached computer), and possibly even that
> the SAS drives are possibly causing some form of congestion or
> performance bottleneck on their controller for all of the SATA drives,
> or rather, the array isn't very good at mixing both types on the same
> controller and keeping up the expected speeds.

There is a single sas controller, with a single sas cable to a single drive 
bay enclosure with 12 bays.

I always use software raid. I am only mirroring sas-to-sas or sata-to-sata, 
matched drives. In fact all 8 of the sata drives are the exact same model, 
from the same shipment.

It is an lsi sas controller, lspci shows:

SCSI storage controller: LSI Logic / Symbios Logic SAS1068 PCI-X Fusion-MPT 
SAS (rev 01)

The sata drives are 320GB 7200rpm hitachi deskstar.

Does any of this info help shed insight to my puzzle?

All of the drives are idle--none are in service yet. It makes no sense to me 
why I would see such significantly different performance characteristics 
based solely on the specific pairing of drives. Perhaps a bios bug in the sas 
controller card?

Something to do with how the addressing is done with sas expanders?