[c-nsp] IPv6 over GRE or IPv6IP tunneling on sup32
Michael Ulitskiy
mulitskiy at acedsl.com
Tue May 25 12:18:01 EDT 2010
On Tuesday 25 May 2010 01:48:02 am Mikael Abrahamsson wrote:
> On Mon, 24 May 2010, Michael Ulitskiy wrote:
>
> > Hello,
> >
> > Does anybody know if sup32/PFC3B support hardware forwarding of ipv6 over GRE or ipv6ip tunnels?
> > IOS 12.2(33)SXH4. I can't find a definitive answer in cisco doc.
> > I know "tunnel mode gre" runs in hardware, but I'm not sure if ipv6 on top of it will make any difference.
> > Has anybody tried it?
>
> I know for a fact that "tunnel mode ipv6ip" is hw path on a Sup720 (if
> it's the only tunnel using that loopback IP as tunnel IP), so should be on
> sup32 as well?
>
My understanding is that sup32 should be no differenct.
Anyway it seems you're right.
I've made the following little lab - Tu0 is IPv6 over GRE tunnel and Tu1 is IPv6IP.
interface Tunnel0
no ip address
load-interval 30
ipv6 address xxxx:yyyy:FFFE:1::1/64
ipv6 enable
no snmp trap link-status
tunnel source Loopback0
tunnel destination a.b.c.d
!
interface Tunnel1
no ip address
load-interval 30
ipv6 address xxxx:yyyy:FFFE:2::1/64
ipv6 enable
no snmp trap link-status
tunnel source Loopback1
tunnel destination a.b.c.d
tunnel mode ipv6ip
!
Here's what I see:
CORE1#sh mls cef ipv6
Codes: + - Push label
Index Prefix Adjacency
...
196864 xxxx:yyyy:FFFE:1::/64 punt
196866 xxxx:yyyy:FFFE:2::/64 Tu1 a.b.c.d
...
GRE tunnel (Tu0) has punt adjacency, while ipv6ip tunnel has Tu1 adjacency.
Also I've put some traffic on both tunnels and here's what I got:
CORE1#sh int tu0 stats
Tunnel0
Switching path Pkts In Chars In Pkts Out Chars Out
Processor 3 264 3 408
Route cache 4212 5677980 27110 25666796
Distributed cache 10152 6065737 0 0
Total 14367 11743981 27113 25667204
CORE1#sh int tu1 stats
Tunnel1
Switching path Pkts In Chars In Pkts Out Chars Out
Processor 9 1078 58 5680
Route cache 0 0 0 0
Distributed cache 29935 29559402 21701 20714008
Total 29944 29560480 21759 20719688
Most ipv6 over GRE traffic is in route cache path, while most ipv6ip traffic
in distributed cache. I'm not sure what the difference between them, but my
guess is that route cache is software CEF, while distributed cache is PFC3.
Am I right?
Is there any better way to confirm it?
Thanks,
Michael
More information about the cisco-nsp
mailing list