Re: MPLS Route Targets

Paul Negron @ 25/8/2010 11:16 -0300 dixit:
> I can do nothing for you if you do not concede that RT's identify the VPN.
> (Not the RD). If this is true, you have fallen into a state of Ignorance. I

No, that is not true. Read all the thread, I have never said that.
(you are putting your interpretation of my words in my mouth,
or my fingers :)

I've never attached a VPN identity. In fact, given the flexibility of
MPLS based VPNs, it is very hard to have a generic definition of
a VPN id.

I have always talked about *route* identity. And I guess we can
agree that a route and a vpn are different things.

> am sure ANY CCIE-SP will correct you on that point. You claimed you didn't
> mind being wrong. I do not believe that is true for 1 second.

You insist in bringing non technical things into the thread.

> I just don't get how you can say you have little experience with Inter AS
> VPN and you come out to correct me after I have used and deployed it for
> years. I am not saying that I cannot be proven wrong but I know I would be
> taking a different approach then you. I guess I could turn it around by
> stating:
>
> You show me ONE instance where the RD has to match and I will believe you. I
> am tired of proving your statement wrong. You prove it right Sir.

Fair. I'll work on this. Basically, I need a situation where I need the
central policy to override the PE policy. Weird, I admit.
But I've seen many here admit that CCIE lab test is a test of
understanding knobs even if used in non standar (or reasonable) ways.
Given that... it's not hard to assemble something along the lines:

Given that you can control the policy on a route reflector, but
not on the edge PEs, make sure that the route is chosen according
YOUR rules and not the ones set by the final PE administrator.

Makes sense in a real world scenario? I do not think so. I take it
that you have never encountered a case that needs this and you have
experience. I do not.

>
> You keep changing your point. I know you don't think so, but EVERYBODY is
> watching. I do not have to judge your Instructing skills, everyone else will
> be judging you. You have done that to yourself.

Yes, I have changed from "same network should use same RD" to
"same network may use same RD", and have learned more about detailed
workings of the BGP selection process and VRF interworking.
But I don't keep changing. It was just an "aha" time when I discovered
how multiple vpnv4 routes are choped into the destination VRF
using the same policy that applies to vpnv4 selection, and
immediatelly back into vpnv4 via route targets, possibly under a
different RD. It behaves just like a redistribution, or at least
that is my view.

>
> Send the dynamips file to me so I can judge for myself concerning your point
> that RD's should match.

-- 
Carlos G Mendioroz  <tron_at_huapi.ba.ar>  LW7 EQI  Argentina
Blogs and organic groups at http://www.ccie.net