Date: Tue, 28 Sep 2010 14:52:54 +0200 From: Ivan Voras <ivoras@freebsd.org> To: freebsd-performance@freebsd.org Subject: Re: gstripe small transaction size Message-ID: <i7soet$o9c$1@dough.gmane.org> In-Reply-To: <4CA1E33C.70200@softhammer.net> References: <4CA14009.4050906@softhammer.net> <i7sguk$gi2$1@dough.gmane.org> <4CA1E33C.70200@softhammer.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On 09/28/10 14:44, Stephen Sanders wrote: > Increasing MAXPHYS and turning up the stripe size won't have the effect > I'm looking for ? I've missed you tuned MAXPHYS up. Yes, in this case it should work, if the underlying driver supports larger IO sizes. As I see it, if all of these are true: * MAXPHYS is upped to be at least stripe_size * number_of_drives (and possibly 2x that, to use "fast mode" of gstripe) * vfs.hirunningspace is > MAXPHYS * the driver supports IOs of at least your stripe_size * the file system doesn't introduce unnecessary choppiness in IO (i.e. use soft-updates) it is reasonable to expect streaming IO to your array will use larger transactions.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?i7soet$o9c$1>