<< | Thread Index | >> ]    [ << | Date Index | >> ]

Subject: Re: bridge + cipe
From: Peter van den Heuvel <peter,AT,asylum,DOT,xs4all,DOT,nl>
Date: Sat, 14 Jul 2001 12:48:53 +0200
In-reply-to: <178859BD4001D511A89D00508BB0D72017F906@WTPS0017>


I think you've posed yourself too big a challenge in one giant leap.
Take more time and smaller steps. If you dont' know how data gets from 
the cipe interface to the ethernet card, you should experiment for a bit
with singel components before you tie 'm all together. In all truth, it
IS complex.

So, make sure you get plain routing to work. Then make sure you get cipe
tunneling to work. Then the radio interface. Then change from routing to
switching / bridging. And finally tie it together.

OK, if you connect two private networks over cipe, you need four
networks: private in location A, private in location B, the Internet (or
whatever insecure net space) and the (point-to-point) network that is
cipe. Every network needs at least one interface on a server. Data moves
from one interface to another by (normally) means of routing. You simply
need the "Internet" as a means for cipe to move data from A to B.
Otherwise you can just ignore it for now. Conceptually the cipe network
is your true external network. It is also the one that your servers
actively route to and from themselves.

Routing works by means of pools of continues IP addresses and some
additional rules that describe what other machine should be contacted to
send data to if the interface is not local (the gateways). Switching
works on the basis of hardware MAC addresses and fast hash tables for
all individials NIC's within the WAN to link these to local network
interfaces. Cause I like to prevent WAN switching (let alone bridging)
(cause of security, bandwith) I've never used cipe in that context.
Because the routing part is not integral to cipe, I bet you could easily
replace it with a switching mechanism. Nevertheless, I bet you must
assign your cipe interfaces their own MAC adrresses in order to get
switching done.

Hope this helps a bit.

> However, I discovered that my CIPE config was inaccurate.  For some reason I
> cannot get the ciped-db daemon to run now.  When I did get it to start my
> options file was not right.  I was pointing my 'ipaddr' and 'me' addresses
> to the same address.  I'm realizing that I don't have as good of a grasp on
> this as I need.
> Question:
> When I specify the 'me' address in the options file is that what takes
> traffic from the cipe interface (cipdb0) and puts it on the "WAN" interface
> (eth1)?  I have been having trouble understanding how data gets from cipdb0
> to whatever physical (eth1) interface I am wanting to use for the cipe link.
> This is the only place I can see that tells the computer what to do with
> cipdb0 traffic.

<< | Thread Index | >> ]    [ << | Date Index | >> ]