On Wed, 20 Mar 2002, Doug McLaughlin wrote:
> Poly-team,
>
> it seems that polymix-4 has a routed configuration that could support our
> examination of the cisco cache engine (transparently, using wccp to
> redirect).
>
> i read that this has been done in the past, but there doesn't seem to be any
> recent work in this area (that i can find posted).
>
> my concern is that the clients & servers will be on two different networks.
>
> if i modify the polymix-4.pg configuration for:
>
> server_side = {
> addr_space = [ 'lo0::10.101.128-251.1-250:80/22' ];
> hosts = [ '172.16.101.191-192' ]; // replace with IPs of your PCs
> };
>
> to reflect the "server" network configuration - should i be set?
Hi Doug,
You probably will need to modify the PGL file when using a router,
instead of a switch. I recently ran some polymix-4 tests with
WCCP. Here's what I did:
I took the normal 172.16.101.0/24 network and divided it into
four subnets:
clients: 172.16.101.0/26
proxy: 172.16.101.64/26
servers: 172.16.101.128/26
I gave each router interface the first address in the subnet (.1,
.65, .129).
I gave the polygraph client, server, and proxy the second address
in each subnet. For example:
client: 172.16.101.2
proxy: 172.16.101.66
server: 172.16.101.130
I left the monitoring PC at .100 and connected it to the proxy
subnet.
The final step is to make the routing work. If you've read our
howto document for the polymix-4 test, you've probably seen the
little shell script that creates routes on the polygraph clients
and servers. Instead of using that, you need to have the clients
and servers (and proxy) forward all packets to the corresponding
router address. Then configure the router with routes to the
10.101.X.0/22 subnets. For example:
ip route 10.101.0.0 255.255.252.0 172.16.101.2
ip route 10.101.128.0 255.255.252.0 172.16.101.130
(I'm not sure about that IOS syntax).
In my setup the polygraph client and server boxes still bind the
10-net addresses to loopback, so they also "route" packets from
their external (i.e., fxp0) interface to the loopback interface.
Let me know if that doesn't help.
This archive was generated by hypermail 2b29 : Mon Feb 06 2006 - 12:00:21 MST