Re: FTP benchmark proposal
Matthew Wilcox (Matthew.Wilcox@genedata.com)
Mon, 28 Jun 1999 12:32:20 +0200
On Mon, Jun 28, 1999 at 04:42:59PM +0800, David Luyer wrote:
> > Larry McVoy wrote:
> > > So I wouldn't worry too much about making the test realistic. if you
> > > can set up a work load that has 6000 sockets going at the server in
> > > parallel, I suspect that it will stress the server just fine. It's when
> > > you try to do the load through a few sockets that all the timing enters
> > > into the equation. Yeah, I'm sure the tests will have to be played
> > > with a bit, but the first step is to just do it and see what we get.
> > > I'll call VA tomorrow and see if they are interested. Red Hat is also
> > > setting up such a lab on the East Coast.
> >
> > of course, with 6000 connections you can't just use one ftp daemon
> > per connection.
>
> I thought the 4192 process limit was gone with the latest kernels?
>
> Granted, you'd probably want to be using an ftp daemon which is quite efficient
> (eg, internalizes 'ls'), and probably want to run a front-end cache which
> serves multiple connections if possible, but doing it with 6000 processes
> would be an interesting test of how Linux could handle such things. (hmm,
> hopefully an ftp daemon which doesn't touch too much RAM in every process
> too...)
Sounds rather like hoser ftpd, brought to you by Zach `phhttpd' Brown
(and others). There was a WIP on this at LinuxExpo and you can find it
at http://www.zabbo.net/hftpd/
--
Matthew Wilcox <willy@bofh.ai>
"Windows and MacOS are products, contrived by engineers in the service of
specific companies. Unix, by contrast, is not so much a product as it is a
painstakingly compiled oral history of the hacker subculture." - N Stephenson
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/