>Is it possible to cause the IP precedence of a GRE packet to be the same as
>the IP precedence of the packet which it encapsulates?
A very interesting problem. Without researching it, I'd be inclined
to say no on a Cisco router, unless you terminate the tunnel at each
router hop, and then set precedence with a route map as you
re-encapsulate it. In formal terms, you have to define the desired
per-hop behavior (PHB) at each hop rather than end-to-end.
There might be a really kludgy way to do it with BGP QoS Policy
Propagation, if the destination addresses were different.
You MIGHT be able to do it on a Bay/Nortel RS router, because it does
allow the equivalent of access lists on pure byte strings. I _think_
the string match is long enough to reach the second IP header. Even
if I could, I don't think I would.
>
>I have a client who is passing real-time as well as normal data over a 3DES
>encrypted tunnel. I have had to resort to using separate
>tunnels for the two data streams, but I consider this to be a sub-optimal
>solution.
But here's where I start to question. Why do you consider two
tunnels to be suboptimal? I'd say the general consensus among MPLS
traffic engineering people is to associate a priority with a tunnel,
merge different flows of the same priority onto what is now called a
"traffic trunk", then put the multipriority traffic back together at
the egress. It's MUCH easier to do traffic engineering when the
tunnel/trunk has a single priority. Remember that traffic
engineering implies the reservation of bandwidth.
>
>For reference, I am using a 2621 at one end and a 3640 at the other with
>12.1.5 images.
>
>This is a real world problem for me. Would this kind of thing possibly come
>up on the CCIE R/S exams?
>
>Chris Read
Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=19131&t=19125
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
