[c-nsp] Basic understanding of 6PE and 6VPE

adam vitkovsky adam.vitkovsky at swan.sk
Mon Jun 25 10:37:55 EDT 2012


If the MPLS backbone doesn't have the TTL propagation enabled than the ICMP
can fail only on Ingres PE or on Egress PE 
If you trace from San Fran to NY do you really need to see all the P-core
nodes along the path sitting there not caring about IP at all just switching
labels all day?
In the old days when there was a frame-relay pvc between the two routers we
didn't see the underlying FR network either

And what about VZ or L3 do they have their global internetworks configured
with native IPv6 please?

I could imagine this possible by configuring only subset of the P-core nodes
and their interconnecting links with native IPv6 just enough for the major
POPs to be interconnected maybe even for the cost of suboptimal path


adam
-----Original Message-----
From: Jared Mauch [mailto:jared at puck.nether.net] 
Sent: Monday, June 25, 2012 3:35 PM
To: adam vitkovsky
Cc: 'Mikael Abrahamsson'; 'Oliver Boehmer (oboehmer)';
cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] Basic understanding of 6PE and 6VPE

Adam,

you are missing the fact that you may have ICMPv6 error generation be
suppressed or dropped.  Traceroute and other things are valuable tools to
diagnose what is going on.  Having an indicator of where the device sits
that generates the error within the "MPLS Core" is important.

Those that do loose-rpf on IPv6 may also see the mapped-v4 addresses
dropped.  This is observed by tracing across the ATT and Cogent MPLS network
as they have done 6PE.

If you want to repeat this problem/mistake then go ahead.  I'm trying to
help folks understand why I'm making the recommendation.

- Jared

On Jun 25, 2012, at 9:30 AM, adam vitkovsky wrote:

> But that might require lot of work like configuring all the P-core 
> routers with BGP and update all the RRs with additional sessions to 
> P-core routers and update BGP on all ASBRs with AFI 2 I still fail to 
> see the added value if the same service can be achieved with BGP free 
> core
> 
> adam
> -----Original Message-----
> From: Mikael Abrahamsson [mailto:swmike at swm.pp.se]
> Sent: Monday, June 25, 2012 2:40 PM
> To: adam vitkovsky
> Cc: 'Jared Mauch'; 'Oliver Boehmer (oboehmer)'; 
> cisco-nsp at puck.nether.net
> Subject: Re: [c-nsp] Basic understanding of 6PE and 6VPE
> 
> On Mon, 25 Jun 2012, adam vitkovsky wrote:
> 
>> If you run 6PE on L2TPv3 core than maybe, otherwise I don't see any 
>> added value in running IPv6 in the MPLS core
> 
> The added value is that you have the choice of not labelswitching IPv6 
> at all, ie running it native.
> 
> -- 
> Mikael Abrahamsson    email: swmike at swm.pp.se




More information about the cisco-nsp mailing list