From: Yves Fauser (Yves@xxxxxxxxx)
Date: Mon Aug 13 2001 - 09:56:45 GMT-3
Hi all,
jonatale@earthlink.net asked me if an isis clns paket does have a TTL field, so
I
read some a bit more and came to this conclusion :
Isis doesn't use clns at all, isis used it's own packet format (isis PDU, unlik
e
ospf that uses an IP packet). clns has a lifetime field, which is the same as a
TTL. The isis PDU doesn't have a lifetime field. Isis PDU's also don't have a
Layer 3 multicast destination address, they only have a MAC multicast address o
f
0180.C200.0015. From my understanding a FR HUB router with a physical or
multipoint interface will never "replicate" a isis PDU from Spoke to Spoke. Als
o
it is the general rule for a hello packet to find neighbors. A spoke will never
be
a neighbor of another spoke regardless what you do.
jonatale@earthlink.net also came up with the idea of creating a tunnel between
the
two spokes, this works fine.
Yves
Yves Fauser wrote:
> Bryan,
>
> I think you are right, we may both have some kind of a IOS Version trouble,
> but more likely it will not work on Point-to-Multipoint or Physical FR
> interfaces in any Version. I saw a scenario that worked with isis on FR PtM
> interfaces (in ECP1), but this scenario had a major difference, one spoke was
> a level-1 and the other spoke was a level-2 router of another Area. I just
> tried it out again, and it works in my home lab. After reading some more of
> Doyle VolI I think I know why (please correct me if I'm wrong guys).
>
> Doyle, Page 607 :
> "Unlike OSPF ISIS router attached to a broadcast multi-access network
> establishes adjacencies with all of its neighbors on the network, not just th
e
> DR. Each Router multicasts its LSP's to all of his neighbors, and the DR uses
> a system of PDU's called Sequence number PDU's (SNP) to ensure that flooding
> is reliable."
>
> Doyle, Page 608 :
> "As the L1 and L2 priorities suggest, separate DR's are elected on a network
> for level 1 and level 2."
>
> In OSPF the DR is like a "route reflector" in BGP. In ISIS the DR is only use
d
> to control the Exchange of LSP's between the neighbors. Since the ISIS and
> OSPF multicast/unicast will have a TTL of 1, they will die on the FR HUB. In
> OSPF, when the DR is the HUB, there is no need for the LSA to take 2 hops
> (Reflection). In ISIS it has to take 2 hops to work, and this is not possible
> since the TTL is 1. If you have 2 Spokes, one is a level-1 and the other is a
> level-2 it works, since two separate adjacencies are build. If both are
> level-1 or both are level-2, it will not work since the DR concept of ISIS ha
s
> no knowledge of NBMA Networks. If you have 3 spokes you can't use this trick
> anymore since you must have either 2 level-1 1 level-2 routers ore the
> opposite.
>
> Any comments are more than welcome,
>
> Good luck, Yves
>
> "Cox, Bryan" wrote:
>
> > Group,
> >
> > I have come to the conclusion that ISIS won't run in a frame-relay
> > environment with physical or multipoint interfaces without a full-mesh.
> > Even with "frame map clns <dlci> broadcast" on the spokes and hub I never
> > see the ISIS hellos being received at a spoke from another spoke. Thus no
> > adjacency is formed between the spokes.
> >
> > In a full mesh I see the appropriate hellos between all points in the
> > frame-relay network. IS-IS forms adjacencies and all routes are installed
> > in the table.
> >
> > Does anybody have any working configs for a hub and spoke environment with
> > point-to-multipoint or physical interfaces that they can post that say
> > otherswise? Or am I on the right track?
> >
> > Thanks,
> >
> > Bryan
> > San Jose October 25th
> > **Please read:http://www.groupstudy.com/list/posting.html
> **Please read:http://www.groupstudy.com/list/posting.html
**Please read:http://www.groupstudy.com/list/posting.html
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:31:50 GMT-3