[4017] in linux-net channel archive
Re: Virtual web server
daemon@ATHENA.MIT.EDU (Michael Slater)
Thu Aug 15 08:40:59 1996
Date: Wed, 14 Aug 1996 08:55:02 +0800 (WST)
From: Michael Slater <aonline@sol.tnet.com.au>
To: Al Longyear <longyear@netcom.com>
cc: linux-net@vger.rutgers.edu
In-Reply-To: <199608120442.VAA25430@netcom22.netcom.com>
On Sun, 11 Aug 1996, Al Longyear wrote:
> I have a small problem.
>
> I have one customer who has a single computer system upon which he
> wants to offer virtual web services.
>
> Using dummy device(s) will work for a small number.
>
> However, this customer wants to dedicate the entire class-C network to
> this one computer for web services. This is ONE computer and 255
> IP addresses -- all on one network -- all on one computer.
>
> Routing is not a problem. We just make the system the gateway to the
> class-C network and the frames are delivered to this system as they
> would be for any router.
>
> However, does anyone have a way to 'properly' define that all 255
> possible IP addresses for a class-C network are local and not to be
> forwarded to another network?
>
> I can hack the networking code to do this -- even if I must put
> specific recognition of the IP network into the networking
> code. However, that is not a good solution. (Other than a definition
> of 'good' meaning "it works!")
>
> This is not a problem of the computer being on two networks. This is a
> problem of the entire network BEING that computer. If it was just the
> one computer on two networks then either a single dummy or aliases
> would do.
>
> I need to have the destination IP address intact in order to do the
> virtual services in apache.
>
> Has anyone a better idea as to the solution which will work for this
> problem? I have tried several ideas. They all did not work -- short
> of doing a 255 dummy devices module and giving each one a specific
> entry from the IP network.
Im not an expert, but whats wrong with doing it that way ?
>
> I have tried to use IP alias devices. They do the same thing with the
> same set of limitations.
>
> Many thanks.
>
> --
> Al Longyear longyear@netcom.com
> Finger for PGP key
>
