Folkert said:
And another one that will be banned from playing in Arnie's sandbox.
Uhuh.
You think that 20+ years of existence is not enough to learn how SCSI works.
I wonder what happens if you apply this chain of thought to RAID.
Actually, applying your chain of thought it should be tremendously more sound
than SATA just because it has been around for so much longer than SATA.
No, thats not the case, because people aren't running with equipment
and operating systems that are 20 years old. New SCSI MBs and
controllers require new and ongoing support as do other hardware
choices. The more samples you have the more likely that you will have
found all of the problems. There are far more samples for any given
SATA subsystem than there is for any given SCSI subsystem. People have
been using RS-232 serial communications for much longer than USB, so
does that mean that RS-232 devices are more reliable than USB devices?
Of course not; that's just plain stupid.
The most "reliable" system is one with a large single disk, and SCSI is
not a higher performance solution on a single disk system. A multiple
disk system has more parts to fail, has more layers of software to have
bugs, draws more power and generates more heat. SCSIs performance
advantages are only with multiple-disk systems. We use large Sata disks
and mirror them with rsync. Inexpensive, reliable and fast (enough).
Most people who've never done any actual testing have wrong ideas about
just about everything. Your logic is the same logic that says that a
system with 2 cpus must be faster than a system with 1 cpu. It seems
logical, but its simply not necessarily the case, and often isn't the
case at all.
I also find it comical that the same clan of folks who whine about the
power consumption advantages of AMD processors have no like opinions
when it comes to other equipment. Its a lot like politics; the case
that fits whatever it is you're doing is the one that suits you for the
present conversation.
TM