[Gluster-devel] Striping.

[Anand Avati]

Yes, also you should be using read-ahead on the client side to leverage read
throughput with stripe (else it would end up serial alternative reads). Also
write-behind should be used for acheiving the aggregated disk throughput
with stripe.

avati

2007/11/29, Chris Johnson <johnson at nmr.mgh.harvard.edu>:
>
> On Thu, 29 Nov 2007, Anand Avati wrote:
>
>       So what I'm hearing is that my test iss too simple.  That right?
>
> > Chris,
> > you will see the help of stripe when you perform IO where disk head is
> the
> > bottleneck. For example try to dd a file > 2x RAM size. Then dd with
> stripe.
> > you will note that the 'rock bottom throughput' is lifted higher with
> > stripe. It also helps when one file is accessed by a LOT of clients so
> that
> > one machine's IO does not become a bottleneck.
> >
> > avati
> >
> > 2007/11/29, Chris Johnson <johnson at nmr.mgh.harvard.edu>:
> >>
> >>       Hi again,
> >>
> >>       So I did my dd read test of my 24MB file on a simple two brick
> >> stripe.  Nothing fancy.  Two bricks and the stripe define on the server
> >> side.  Client side mounts the remote stripe.  Default block size,
> >> 128KB.
> >>
> >>       No difference between that and my single brick test as far as
> >> time goes.  What's striping supposed to get me?  I thought it was
> >> drives heads.  Or am I missing something again?
> >>
> >>
> -------------------------------------------------------------------------------
> >>
> >> Chris Johnson               |Internet: johnson at nmr.mgh.harvard.edu
> >> Systems Administrator       |Web:
> http://www.nmr.mgh.harvard.edu/~johnson
> >> <http://www.nmr.mgh.harvard.edu/%7Ejohnson>
> >> NMR Center                  |Voice:    617.726.0949
> >> Mass. General Hospital      |FAX:      617.726.7422
> >> 149 (2301) 13th Street      |Survival, all by it self, isn't worth it.
> >> Charlestown, MA., 02129 USA |                 Me
> >>
> >>
> -------------------------------------------------------------------------------
> >>
> >>
> >> _______________________________________________
> >> Gluster-devel mailing list
> >> Gluster-devel at nongnu.org
> >> http://lists.nongnu.org/mailman/listinfo/gluster-devel
> >>
> >
> >
> >
> > --
> > It always takes longer than you expect, even when you take into account
> > Hofstadter's Law.
> >
> > -- Hofstadter's Law
> >
>
>
> -------------------------------------------------------------------------------
> Chris Johnson               |Internet: johnson at nmr.mgh.harvard.edu
> Systems Administrator       |Web:
> http://www.nmr.mgh.harvard.edu/~johnson
> NMR Center                  |Voice:    617.726.0949
> Mass. General Hospital      |FAX:      617.726.7422
> 149 (2301) 13th Street      |What the country needs is dirtier fingernails
> and
> Charlestown, MA., 02129 USA |cleaner minds.  Will Rogers
>
> -------------------------------------------------------------------------------
>



-- 
It always takes longer than you expect, even when you take into account
Hofstadter's Law.

-- Hofstadter's Law