OT: external SATA RAID

Eberhard Moenkeberg emoenke at gwdg.de
Tue May 11 19:09:01 CDT 2004


On Tue, 11 May 2004, Eric Swenson wrote:

> I've heard rumors that the next generation of external direct-attach
> SerialATA-SCSI raid arrays will have ~120 Mbyte/sec performance, but
> I've yet to see real benchmarks of them.

I can produce this as a peak value already with my "old" masscope
array - f.e. by copying a loop-mounted ISO image without any
concurrent activity. To transport this rate over the net would be
the harder task...

I still have not tested the masscope+ (S-ATA drives), but I guess it is
not a better performance class than the "old" (P-ATA) masscope because
the CPU seems to be equal (PowerPC G3 CPU 400 MHz), and maybe some more
parts of the controller design - f.e. two internal 64-bit PCI busses and
"single disk per channel" design.

Today, a single S-ATA drive is equal in performance with a single P-ATA
drive. This may vary with tomorrow's drives which may run faster than
7200 rpm and may have bigger caches than 8 MB and may win some percent
from the future S-ATA 2.0 feature "tagged command queuing" (or not - a
clever firmware can emulate it, just like the linux kernel does), but at
current time I consider all those "better!" exclamations as pure marketing
gags as long as they belong to "inside" features of "black boxes" only.

If I take the "old" masscope as a black box with a 3-year warranty (even
longer if I pay some more percent), why should I choose a masscope+?
The hot-plug and the cooling design seems good enough, so I see no
benefit choosing S-ATA currently, as long as the difference is inside the
black box only, with identical characteristics "outside".

Cheers -e
Eberhard Moenkeberg (emoenke at gwdg.de, em at kki.org)

More information about the Linux-PowerEdge mailing list