may be the linux ramdisk is another good choice.
you can use memory like normal filesystem.

----- Original Message ----- 
From: "Frank Wiles" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Cc: <modperl@perl.apache.org>
Sent: Wednesday, September 27, 2006 9:45 PM
Subject: Re: mod_perl using IPC::ShareLite


> On Wed, 27 Sep 2006 13:13:51 +0100
> "Ben Wilder" <[EMAIL PROTECTED]> wrote:
> 
> > My research so far has led me to believe that IPC::ShareLite could be
> > used by the cron job script to set a few (very small) variables into
> > a portion of shared memory, that a separate mod_perl application
> > could poll. Mod_perl would not need to modify these variables, only
> > read them.
> > 
> > Would using a library such as IPC::ShareLite be considered a fudge
> > for this type of action, is there some sort of global apache variable
> > section that I can modify with one application and read out from
> > another?
> > 
> > Any thoughts appreciated!
>  
>    Using one of the IPC modules is definitely a reasonable choice
>    based on what you're trying to do.  But I wanted to throw out
>    some other options as some people find IPC hard to work with. 
> 
>    1. BerkeleyDB or sqlite.  I realize you said you'd like to avoid a
>        database, but sometimes this means avoid installing a big
>        RDMS like PostgreSQL, MySQL, etc. 
>  
>    2. memcached and Cache::Memcached could store your data in
>        a memory cache for you. 
> 
>    3.  Use can also use a simple config file with something like
>         Config::General.  Your crons can update this data onto disk
>         and your mod_perl apps can simply read it in as a config file,
>         maybe using a short 5 minute cache to reduce disk reads. 
> 
>    4.  You could also use Storable to drop an object full of your
>         small variables from your cron and have your mod_perl app
>         read them in on each request ( or a short cache in pnotes or
>         something ). 
> 
>    The last two may sound slow due to the disk reads, but it really
>    shouldn't cause a performance problem in all but the most 
>    demanding environments. 
> 
>  ---------------------------------
>    Frank Wiles <[EMAIL PROTECTED]>
>    http://www.wiles.org
>  ---------------------------------
> 

Reply via email to