this is something i have wondered also. ive never seen cpu go over
20-25% on a resync. all the raid controllers i've seen (even the
cheapest) allow you to specify the percentage of cpu to devote to the
rebuild.
----------------------------------------------------------------------
WARNING, everything below this line may be a complete waste of your
time and not worth reading.
on a more absurd note, and please forgive me for venting about topics
everyone has heard too many times already...
i realize that benchmarks are mostly bunk, but one thing i would be
interested to see, just out of curiosity is this:
a modern lvd or 160 system running linux or freebsd, apache, php,
mysql, software raid5
and
another machine, same hardware spec running w2k, IIS, asp, sql7,
software raid5
put a deep and fat hierarchy of porn on each raid. have each page do
a database lookup or two.
then run an http benchmarking program that will draw random urls from
a set that refers to every page in the porn tree. once one of the
machines appears to hit its limit, do the following on each machine:
launch 30 heavy perl programs in the background.
mount and unmount some network based filesystems.
install a few drivers.
partition and format another disk.
while this test is anything but scientific i think it will illustrate
interesting differences between the two OS models. one of the reasons
i think raw benchmarks arent all that important is that they dont
come close to describing real world conditions.
who would care if software raid was faster than hardware raid if it
used 90% of the cpu on a GHz Xeon? who would care if w2k raid was
faster than linux raid if the OS cratered under high load?
-jacob
On Mon, Sep 18, 2000 at 04:24:05PM -0700, Dan Jones wrote:
> An interesting comparison between linux Software RAID and
> Windows 2000 Software RAID, focusing on Quality of Service,
> by the Berkeley ISTORE group. The most significant difference
> is described as one of recovery philosophy. Is that really
> hardwired in the software RAIDs? It is a nice feature to
> enable the user to prioritize the recovery rate.
>
> Although not explicitly stated, it appears the linux/apache
> system runs about 10% faster than Win2000/IIS under their
> benchmark.
>
> http://istore.cs.berkeley.edu/talks/iram-jan00-retreat-amebench/sld001.htm
>
> Dan
> -
> To unsubscribe from this list: send the line "unsubscribe linux-raid" in
> the body of a message to [EMAIL PROTECTED]
>
-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to [EMAIL PROTECTED]
