You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@tomcat.apache.org by Remy Maucherat <re...@apache.org> on 2002/11/25 10:58:00 UTC

Optimizing Coyote

Hi,

Many optimizations in Coyote HTTP/1.1 are quite straightforward, but 
some others are more difficult.

Right now, I'm "stuck" trying to optimize the amount of get/setSoTimeout 
method calls. Each of them is relatively expensive, and probably ends up 
exercizing the network stack more than we would like. I will try to 
experiment and commit things. Can you review them Bill ?

Thanks,
Remy


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: Optimizing Coyote

Posted by Remy Maucherat <re...@apache.org>.
Bill Barker wrote:
> At least with the Solaris TCP stack, the SoTimeouts make no difference at
> all to the 'ab' times.  However, it's easy enough to make it a configuration
> option.

Thanks for doing it :)
I don't know if it ends up impacting actual benchmarks (although it may 
depend on the platform, as it ends up with a native call), but it shows 
up in the CPU profiler with OptimizeIt, and I thought it wouldn't hurt 
to try to improve this.

Remy


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: Optimizing Coyote

Posted by Bill Barker <wb...@wilshire.com>.
At least with the Solaris TCP stack, the SoTimeouts make no difference at
all to the 'ab' times.  However, it's easy enough to make it a configuration
option.

----- Original Message -----
From: "Bill Barker" <wb...@wilshire.com>
To: "Tomcat Developers List" <to...@jakarta.apache.org>
Sent: Monday, November 25, 2002 11:54 AM
Subject: Re: Optimizing Coyote


> I'm surprised at the expense:  I had thought that this should be little
more
> than setting a note in the O/S kernel.  It's also pretty much what
> Apache/httpd 2.0 does, at least last time I looked.
>
> However, I'll take another look at them.
>
> ----- Original Message -----
> From: "Remy Maucherat" <re...@apache.org>
> To: "Tomcat Developers List" <to...@jakarta.apache.org>
> Sent: Monday, November 25, 2002 1:58 AM
> Subject: Optimizing Coyote
>
>
> > Hi,
> >
> > Many optimizations in Coyote HTTP/1.1 are quite straightforward, but
> > some others are more difficult.
> >
> > Right now, I'm "stuck" trying to optimize the amount of get/setSoTimeout
> > method calls. Each of them is relatively expensive, and probably ends up
> > exercizing the network stack more than we would like. I will try to
> > experiment and commit things. Can you review them Bill ?
> >
> > Thanks,
> > Remy
> >
> >
> > --
> > To unsubscribe, e-mail:
> <ma...@jakarta.apache.org>
> > For additional commands, e-mail:
> <ma...@jakarta.apache.org>
> >
>
>
> --
> To unsubscribe, e-mail:
<ma...@jakarta.apache.org>
> For additional commands, e-mail:
<ma...@jakarta.apache.org>
>


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: Optimizing Coyote

Posted by Remy Maucherat <re...@apache.org>.
Bill Barker wrote:
> I'm surprised at the expense:  I had thought that this should be little more
> than setting a note in the O/S kernel.  It's also pretty much what
> Apache/httpd 2.0 does, at least last time I looked.
> 
> However, I'll take another look at them.

It's not *that* expensive, it's just that it shows up in the profile. In 
the end, all the calls add up, so it should be optimized IMO.
I does a few things, and then ends up using a native method in the end 
(I think that's relatively expensive in Java; I don't know much about 
performance when dealing with native bits though).

Remy


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>


Re: Optimizing Coyote

Posted by Bill Barker <wb...@wilshire.com>.
I'm surprised at the expense:  I had thought that this should be little more
than setting a note in the O/S kernel.  It's also pretty much what
Apache/httpd 2.0 does, at least last time I looked.

However, I'll take another look at them.

----- Original Message -----
From: "Remy Maucherat" <re...@apache.org>
To: "Tomcat Developers List" <to...@jakarta.apache.org>
Sent: Monday, November 25, 2002 1:58 AM
Subject: Optimizing Coyote


> Hi,
>
> Many optimizations in Coyote HTTP/1.1 are quite straightforward, but
> some others are more difficult.
>
> Right now, I'm "stuck" trying to optimize the amount of get/setSoTimeout
> method calls. Each of them is relatively expensive, and probably ends up
> exercizing the network stack more than we would like. I will try to
> experiment and commit things. Can you review them Bill ?
>
> Thanks,
> Remy
>
>
> --
> To unsubscribe, e-mail:
<ma...@jakarta.apache.org>
> For additional commands, e-mail:
<ma...@jakarta.apache.org>
>


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>