[nsp] MPLS & IPv6

[Lars Erik Gullerud]

Yes, THIS is working, however it is unfortunately not what you are
asking about. I merely stated, based on the phrasing of the original
question, that you can do tag-switching (MPLS) and IPv6 on the same
interface, without any hickups that we have experienced.

We are not doing IPv6 over TE tunnels, nor IPv4 for that matter, so I
have no idea what is currently working or not there. We are only using
MPLS to provide Ethernet-over-MPLS services, and all the PE devices for
that are Riverstone gear. The Cisco/Juniper core are all P devices as
far as MPLS is concerned, and do IPv4/IPv6 natively, with no TE.

However, doing IPv6 over EoMPLS tunnels does work just fine, as you
would obviously expect - and is probably the easiest solution right now,
since there are still a bunch of compatibility issues between the
various vendors IPv6/ISISv6 implementations. Which is one of the reasons
we are not actually offering IPv6 to customers yet, we are only running
very limited trials to a few locations.

/leg

On Mon, 2003-02-03 at 16:55, Me wrote:
> Lars, is this actually working?
> 
> We have a similar setup, but no IPv6 traffic will pass when using isis for 
> ipv6, because IPv6 (according to cisco) cannot traverse TE tunnels, and 
> is-is cannot route ipv6 over a different path than it routes ipv4.  I have 
> been told we need to wait for 'multi-topology' is-is for that to work.  I 
> have been using 12.0(21)ST, maybe 23S has this multi-topology is-is?
> 
> 
> On 3 Feb 2003, Lars Erik Gullerud wrote:
> 
> > Date: 03 Feb 2003 10:04:24 +0100
> > From: Lars Erik Gullerud <lerik at nolink.net>
> > To: George Boulescu <george at roedu.net>
> > Cc: cisco-nsp at puck.nether.net
> > Subject: Re: [nsp] MPLS & IPv6
> > 
> > On Mon, 2003-02-03 at 09:08, George Boulescu wrote:
> > > I was wondering if it is possible to have tag-switching and ipv6 enabled
> > > on the same interface.
> > > 
> > > Anyone tried that ?
> > 
> > Yes. We now use it on production GSR's with 12.0(23)S1, and this works
> > just fine, at least on the linecards we have tested this (typically
> > those who can do dCEFv6, which is not yet all cards). Example from an
> > engine 0 single-port GigE on one of our boxes:
> > 
> > interface GigabitEthernet4/0
> >  description <XXX>
> >  mtu 4470
> >  ip address x.x.x.x 255.255.255.252
> >  no ip redirects
> >  no ip directed-broadcast
> >  no ip proxy-arp
> >  ip router isis <XXX>
> >  negotiation auto
> >  mpls traffic-eng tunnels
> >  ipv6 address X:X:X:X:X/64
> >  ipv6 router isis <XXX>
> >  ip rsvp bandwidth 600000 600000
> > 
> > We are doing RSVP-signalled tunnels through these core boxes, for
> > traffic between EoMPLS PE-routers (not Cisco), and native IPv4 and IPv6.
> > Haven't had a single problem yet, apart from the usual with Cisco and
> > IPv6. Not tried this on other platforms beside GSR yet, as Cisco for
> > some reason haven't gotten dCEFv6 for 7500 into 12.0S (although they
> > have it on 12.2T) or (a working) CEFv6 for 7200.
> > 
> > /leg
> > 
> > 
> > _______________________________________________
> > cisco-nsp mailing list  cisco-nsp at puck.nether.net
> > http://puck.nether.net/mailman/listinfo/cisco-nsp
> > archive at http://puck.nether.net/pipermail/cisco-nsp/
> >