You said nothing of using VMWare as a router, you said it was a firewall. And no, any $1000 router isn't sufficient, advanced policy-based routing features that will do what you want are found more expensive routers like the M/T series from juniper and cost $2400 or more.
It actually does both, a proper-configured firewall does route traffic. And incidentally, a $2400 router is essentially a pretty interface over pretty much the same thing I'm using.
You obviously don't know how XML-RPC works or you would understand that it can do what I'm suggesting.
I did a project on it on my last year of University, but I still don't see how it'll solve my problem, unless I want to write a script on the first computer that queries the second computer. In which case, I don't even need XML-RPC, I can just do it by downloading the remote site and displaying it. And I considered doing it that way, but it wouldn't work out for things like images, so I decided against it.
Then you have a very rudimentary understanding of how network load balancing works, while that may be the primary implementation of it, load balancing applications route traffic, which is what you're trying to accomplish, so generic load balancing techniques can be applied to this situation. There isn't anything designed specifically to route a directory of a website to a different web server because that's pretty idiotic to begin with.
They don't really route traffic, they divvy it up. If you are making rules for certain applications/ports to be load-balanced to different places, then you aren't load balancing, you're routing. And I don't care whether it's the directory or a subdomain, both work fine.
Use squid like I said, since you still insist on using multiple virtual servers, it's even more efficient than network-level routing because it dynamically cache's content.
Why would I want the forum cached? And I still don't think using a proxy makes sense.
I suppose you also missed it when I asked for a way to do it at the network layers.
Better policy routing.
All right, it looks like you know what you're doing here, so maybe you can explain this to me. I have two different subdomains on x86labs.org, forum and www. They run on the same port, 80, on different computers on my internal network. The user sends a packet remotely, which arrives at my router. It goes through some chains, eventually getting to the "prerouting" chain in the "nat" table, which is where the routing decision is made. At the moment, I make the decisions like this:
if($protoport =~ m/^([a-zA-Z]+)\/([0-9]*)$/)
{
my $protocol = $1;
my $port = $2;
my $ip = $DMZ_ALLOWED_INCOMING{$protoport};
print " -> NATing external port '$port' on '$protocol' to DMZ ip '$ip'\n";
`$IPTABLES -t nat -A PREROUTING -p $protocol $FROM_INET --dport $port $LOG forwarded: `;
`$IPTABLES -t nat -A PREROUTING -p $protocol $FROM_INET --dport $port -j DNAT --to-destination $ip`;
}
As you can see, I'm identifying it by protocol and port, then making the routing decision (for my current set of firewall rules, written 100% by me, see
rc.firewall). Now, without layer-7 inspection, which I don't think can identify different domains anyways, how would you suggest I make the routing decision?
Even in a general case: you're running some routing software, and those two sites are set up: what do you check to make the routing decision?
