[Beowulf] Software Raid - devolved
alvin at Mail.Linux-Consulting.com
Tue Dec 13 20:25:59 PST 2005
hi ya michael
On Tue, 13 Dec 2005, Michael T. Prinkey wrote:
> Honestly, I am wondering if the Software/Hardware RAID argument has
> devolved to the state of SCSI versus ATA or (heaven forfend) emacs versus
> vi. 8)
but unlike those preferences, software and hardware raid can be measured
and verified by simply changing hw vs sw raid with the same exact
hardware where the only difference is the raid sybsystem
> My experiences with hardware raid have been consistently lack luster over
> 7 years and several generations of hardware.
it's job security .... we all get to go and fix it regularly.. :-)
> My experiences with software
> RAID servers (specifically built for the task) have been largely positive.
ditto ... except for the silly infant mortality of new disks surviving
past the first 30-60 days which i think is highly dependent upon on where
you buy your disks from and how the systems are built
> When I read comments extolling the virtues of hardware RAID solutions, I
> find myself constantly wondering if I could be missing something after
> some many years and many dozens of deployed units.
the part that some folks like or dislike is the ability to monitor
the raid subsystems ...
- monitoring of hw raid is tricky or impossbile while
monitoring of sw raid is "do as you need/want" until you're happy
> To provide numbers, I am really only concerned if the raid array can
> saturate the gigabit line feeding it. On-server performance is pretty
> useless as no work is ever done directly on the RAID servers.
the only useful "preformance data" is in real life ... and not benchmarks
from benchmark programs
> For reading
> data, the server could certainly saturate gigabit...Bonnie on the NFS
> mount gave roughly 85 MB/sec for software RAID5. When we deployed an
> 8-drive RAID5 array using hardware RAID on the SATA 3ware card,
> performance was on the order of 15 MB/sec.
good numbers .. and probably verifiable too with more details like
which disks, which cpu/mem/motherboard, which distro .. :-)
> Clearly, YMMV.
i suspect the raid performance is also extremely dependent on
the network topology and mixture of compute nodes, storage nodes
and users PCs ( especially windoze )
add to that the various NFS parameters to tune for better performance
vs (bad) system defaults that came with the default distro
things have to be tuned to get better performance
More information about the Beowulf