Discuss digest, Vol 1 #1123 - 10 msgs

[richb at pioneer.ci.net (Rich Braun)]

Derek Atkins <warlord at MIT.EDU> wrote:
> Uh, actually, there is a penalty.  With S/W RAID1 the kernel has to
> perform two writes across the PCI/IDE bus ...This extra writing will
> definitely cause a peformance penalty (on writes) for software raid
> that you wont see in a hardware raid.

It's not measurable because the drive speed is well under 1/2 the PCI bus
speed (more like 1/200th).

> Also, historically it had NOT been recommended to use both parts of an
> IDE bus because the master/slave relationship reduces the bus
> throughput.  Has this changed recently?

I'm not talking master/slave.  I'm talking RAID1 on a two-drive configuration:
 one drive as master on each of the two independent IDE buses.  It's been
something like 10 years since any motherboard came out that didn't have
independent IDE buses.

Try a benchmark on a 2-drive software RAID1 vs. a 2-drive hardware RAID1.  You
won't be able to measure the difference.  I tried and couldn't find any.

-rich