On Wed, Sep 19, 2001 at 12:16:56AM +0800 or thereabouts, Juan Miguel Cacho wrote:
> En Tue, Sep 18, 2001 at 08:16:53PM +0800, Rino Mardo escribio:
> #_
> #_ so now that i'm "Linux amnesiac 2.4.9-xfs #15 Tue Sep 18 19:30:09 PHT
> #_ 2001 i686 unknown" i think i better put together a one page mini-HOWTO
> #_ about my xfs experience (or lack thereof). i'm attaching the results
> #_ of my benchmarks using "bonnie". i used "bonnie" for two reasons:
> #_ one, the man page says it's for benchmarking filesystems and two, this
> #_ is my first time to do benchmarks and "bonnie" fits my simple needs.
> #_ maybe when i'm more adept with benchmarking and it's tools i will used
> #_ more fancy ones.
> #_
> #_ so, mr. jijo, now that i've tried all linux filesystems available sans
> #_ "ext3" and benchmarked them maybe i can invite you to try the same on
> #_ your box. ;-)
>
> Interesting, from your benchmark, jfs uses _less_ cpu than the other two,
> and reiser seems to use the most.
>
uh-huh, kaya nga i find them benchmarks subjective. it all depends
ika nga. but overall usage i feel that xfs *is* faster than the other
three specially when i got hit by two power outages here yesterday.
fortunately i was on xfs already. when i booted it's almost like
nothing is being fscked checked but gleaning on dmesg shows xfs
recovering some sessions or logs. i could just imagine if the power
outage hit me when i was using reiserfs or jfs. lalo na ext2!
in the end, personal preferences. i still got some race condition
problem with jfs so xfs muna ako.
--
Waiting for the "Stinky Pete" release of Debian
PGP signature
