Index: [Article Count Order] [Thread]

Date:  Tue, 13 Mar 2007 11:18:10 -0700
From:  "Adam Crews" <adam.crews (at mark)>
Subject:  [coba-e:09143] Re: AW: Re: storing php sessions in memory
To:  coba-e (at mark)
Message-Id:  <1486c6440703131118v31a6c7f1j9869f209d4b1f843 (at mark)>
In-Reply-To:  <027b01c76590$6133cff0$6401a8c0@acerbaby>
References:  <015401c7655a$21c7e940$3701a8c0 (at mark) lapxp>	 <1486c6440703130715q3da7994eh7de2087d89acc59e (at mark)>	 <027b01c76590$6133cff0$6401a8c0 (at mark) acerbaby>
X-Mail-Count: 09143

Also remember, that much of these things are grasping at straws.
Linux does not come configured to be slow, many of the default
settings are balances between the various pros and cons.  If you
really are looking to squeeze more performance, then you probably just
want a bigger box, or more specialized hardware.  I would only expect
minor and probably not even user noticeable improvements after a great
deal of tuning.  It is not until you get specialized applications
where you can really take advantage of certain tuning parameters.


On 3/13/07, Florian Arzberger <arzberger (at mark)> wrote:
> Adam Crews <mailto:adam.crews (at mark)> schrieb am Dienstag, 13. Mz 2007
> 15:16:
> >
> > A big one is to simply mount file systems with the 'noatime' flag.
> > This will save a disk write when a file is accessed.  The FS will
> > still record any modification time, but it will not record if a file
> > is simply read.  On a print server that I run, it resulted in about a
> > 50% speedup in job throughput for large numbers of jobs.
> >
> sounds quite senseful to me. any negative thoughts about it?
> --

----------------------------------------------------------------- Donation Based Web Hosting