[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: good jbod hardware

> > There is no performance-hit with UDMA100/133 drives when using two
> > devices per channel. This an old "SCSI-is-best" story that were true a
> > long time ago.
> On http://www.coker.com.au/hardware/46g.png I have a graph of this using
> ATA66 on an Athlon800 machine without DDR RAM.  It shows a significant
> performance difference between having only one active drive and having
> drives active on different channels.  But the difference between two
> on different channels and two drives on the same channel is much

I get a 404 on the above image. Regardless, you may not notice any big
performance different when running 2 devices per IDE channel/port if you
are doing light random accesses and such, but if you start doing heavy I/O
and such, you'll notice an increasing performance difference compared with
1 devices on each IDE channel. At least, this is our experience.

> ATA133 is twice as fast as ATA66 and modern CPUs are twice as fast as
the one
> I tested with, but you can't buy IDE drives that are twice as fast as
> drives I tested with!  Therefore I expect to see less difference if the
> is repeated on modern hardware.

True. 7200rpm ATA66 VS 7200rpm ATA133 would produce a minimal performance
increase, assuming that the onboard cache and associated caching algorithm
was similar.


To UNSUBSCRIBE, email to debian-isp-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

Reply to: