remember that any bk repository will have all the info in it, so just
access that instead of bkbits.org (yes this will require someone to run
the evil propriatary software ;-), but it will avoid the load on Larry's
server)
David Lang
On Thu, 6 Feb 2003, Willy Tarreau wrote:
> Date: Thu, 6 Feb 2003 07:02:23 +0100
> From: Willy Tarreau <willy@w.ods.org>
> To: Larry McVoy <lm@work.bitmover.com>, Ben Collins <bcollins@debian.org>,
> Larry McVoy <lm@bitmover.com>, Andrea Arcangeli <andrea@e-mind.com>,
> linux-kernel@vger.kernel.org
> Subject: Re: openbkweb-0.0
>
> On Wed, Feb 05, 2003 at 08:37:37PM -0800, Larry McVoy wrote:
> > > You may want to enable mod_deflate, and then scripts can easily make use
> > > of gzip compressed data. May not be an end-all, but something to
> > > consider.
> >
> > Gzip will give 4:1 what these scripts are doing is more like 1000:1.
> > So gzipping the data gets you down to 250:1. That's still way more
> > bandwidth, way too much to be acceptable.
>
> Larry, would it be acceptable/possible to regularly push some data/metadata
> to sites like kernel.org that people already consult for kernel development ?
> This way, Andrea's tool would only have to check kernel.org, and not bkbits.net.
>
> Another solution is to fetch from a reverse proxy-cache on a high-bandwidth
> site, provided that we know what to cache, of course. This could even reduce
> your current HTTP usage since nearly everything should be cacheable for a very
> long period.
>
> You know, sometimes I fetch changesets from bkbits.net with my browser, and
> I later convert them from html to text with a very trivial sed script. And it
> happens that I remember you saying that the bandwidth costs you very much, then
> I feel a bit guilty (although about once a week may not be too much) and I
> wonder what would happen if everyone did the same regularly.
>
> You may also try to cap the bandwidth from the web server, to dissuade people
> from using it, but still not closing it. This could also help you not to pay
> for the extra bytes.
>
> Just some suggestions, of course. I wouldn't like your http service to be
> closed since I sometimes use it.
>
> Cheers,
> Willy
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Fri Feb 07 2003 - 22:00:20 EST