Hi Itojun,

>         in my opinion, site border routers need to have ability to run
>         separate entity of RIP/OSPFv3/IS-IS for each site (don't mix them up).
>         there's no need for protocol modification, since there will be no
>         interaction between routes in site A and site B.

I am not quite sure what you mean...

Assume that I have a router with 4 interfaces (A, B, C and D) in two sites (S1 & S2), 
with
interfaces A & B in S1, and interfaces C & D in site 2, all on an OSPF network.  How 
many 
instances of OSPF would I need to run?

Your message seems to indicate that I would need to run two -- one for S1 and one for 
S2.
But, how would those two instance cooperate to calculate the global routing 
information for 
all four interfaces?

Alternatively, I could run three copies -- one for S1, one for S2 and one for the 
global routes.
But, how would that work?  I'd end up with two copies of OSPF running over each 
interface.
Would they have the same, or different peers?


>         NEC IX router is the only implementation supporting this, as far as
>         i know (i'm a bit embarrassed, KAME doesn't handle this - yet).

Does the NEC implementation use multi-instance routing protocols?  Do you know anyone
on this team?  Could we get them to describe how they handle the site border router 
case
in detail?

I am very interested in understanding how this works.

Thanks,
Margaret

--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to