From: Tim (ccie2be@nyc.rr.com)
Date: Sun Oct 23 2005 - 15:09:53 GMT-3
Steve,
Actually, it's not so hard to know when you need to map the LL address.
Here's what you need to know.
With IPv6, routing protocols use the LL address as the source address when
they send out updates to their neighbors.
That fact is the key fact you have to know and understand.
Now, the router that receives these updates knows on which physical
interface it received the update but what if the physical interface is
sub-divided into multiple logical L3 interfaces as may be the case with f/r
or ATM where multiple PVC's are defined?
In this case, when the receiving router wants to send a packet to a subnet
it learned about over a PVC from a neighbor which PVC should it use?
Remember, the receiving router knows the source LL ipv6 address of the
routing update it received and you can verify that by looking at the ipv6
route table. You'll see the LL ipv6 address as the next hop for the subnets
learned from that router's neighbors.
But, what DLCI should the router use to send packets towards the LL ipv6
address it sees as the next hop?
The router won't know unless you tell it. So, to make a long story short,
if you have an L3 multipoint nbma interface, you need to tell it what dlci
to use to reach a given LL ipv6 address.
HTH, Tim
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
stephen skinner
Sent: Sunday, October 23, 2005 1:42 PM
To: ccielab@groupstudy.com
Subject: IPV6 to link local or not link local , that is the question!
Chaps,
i am trying to collect a list of when a link local IPV6 mapping should be
done .
int s0/0 = NO
int s0/0.1 point-to-point NO
int s0/0.1 point-to-multipoint YES
int s0/0
ip ospf network point-to-point = NO
int s0/0.1 point-to-point
ip ospf network non broadcast = YES
int s0/0.1 multipoint
ip ospf network non broadcast =YES
int s0/0- s0/0.1 p2p - s0/0.1 m2m
ipv6 rip = YES
int s0/0- s0/0.1 p2p - s0/0.1 p2m
ipv6 isis = YES
int s0/0- s0/0.1 p2p - s0/0.1 p2m
ipv6 eigrp = yes
i believe i`m getting a Little confused,
In the labs that i have been doing , some have link local mappings and some
don`t ...
I have been unable to find a proper explanation of when to map Link Local
and
more importantly when not to map Link Local on the CCO .
several examples of ipv6 config on the CCO say that you don`t need to map
LL but further down in the config i see that LL`s are mapped .
one explanation of link local addressing and mapping i have read said
" a dynamic routing protocol need Link Local scope for next-hop ipv6
addressing , therefore create local link mapping`s on you serial int "
yet i have also read that you don`t need it . ( i have also proved it by
labbing this up FR hub and spoke )
so to my QUESTION ..
on the LAB , do you think it is best TO map LL or NOT TO .. if you are not
given any explicit instruction`s to use LL mappings
would anyone PLEASE care to comment
many thanks in advance
Steve
This archive was generated by hypermail 2.1.4 : Sun Nov 06 2005 - 22:00:52 GMT-3