Date: Wed, 21 Jul 1999 10:52:56 -0700 (PDT)
From: Dean Gaudet <dgaudet@arctic.org>
To: new-httpd@apache.org
Subject: Re: [PATCH] 10x performance increase patch #1
In-Reply-To: <199907211728.KAA30422@trudge.engr.sgi.com>
Message-ID: <Pine.LNX.3.96dg4.990721103804.8267E-100000@twinlark.arctic.org>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: new-httpd-owner@apache.org
Precedence: bulk
Reply-To: new-httpd@apache.org


On Wed, 21 Jul 1999, Mike Abbott wrote:

> > > I chose not to port file sizes, bytes written, content lengths, and the
> > > like to 64 bits
> > Can I convince you to join the MPM effort and go to these lengths?
> 
> You'll have to ask my boss :).  The invitation is flattering.  I'm not
> exactly sure what MPM is since I have focused solely on 1.3.x until now,
> but I do hope to join the 2.0 effort soon.  Beyond that I cannot say.

MPM is a playground right now for something that will become one of the
2.x versions.  The main thing which I think will appeal to you is that
we've abstracted http_main -- for example, it would be easy for you to
write an sproc-based MPM (multiprocessing method) which distributes the
load around a zillion processor SGI box in an efficient manner.  Your code
wouldn't really interfere with anyone else's, and you get to benefit from
everyone else's work in modules, and in the protocol core. 

We haven't completely defined what a "thread" is, but it's definately less
heavy than a pthread... we don't use any asynchronous notification, for
example... which gives us the chance to use the lightest possible sproc or
clone based implementation.  I'm not 100% sure, but I think sprocs are
lighter than pthreads on irix... similar to the clone/linuxthreads
relationship.

If you want to take a look at it, grab the latest apache-2.0 snapshot at
<http://dev.apache.org/from-cvs/apache-2.0/>. 

Dean