Hi Mike,

Yes, there are a number of possible ways to address this.

This would be one way to go about it. However, I don't think it is desirable.
Especially when considering that in some cases operators like to advertise
prefixes for local interfaces as external routes on spoke routers. They could
use the R-bit to make the spoke not pass transit traffic. If we chose this
option, then all prefixes would have to be advertised as intra-area prefixes.
Therefore I'm reluctant to use this method.

Regards,
Michael

------ Original Message ------
Received: Mon, 14 May 2012 04:36:26 PM PDT
From: "Mike Dubrovskiy (mdubrovs)" <[email protected]>
To: "Michael Barnes" <[email protected]>, <[email protected]>Cc:
<[email protected]>
Subject: RE: [OSPF] ABR/ASBR with clear R-bit?

> Hi Michael,
> 
> It seems that there's more than one way to skin a cat.
> 
> How about this one:
> 
> Router with R-bit cleared should be area internal router.
> In case if R-bit is cleared on ABR or ASBR, the router must give up the
> ABR/ASBR status.
> 
> Mike
> 
> -----Original Message-----
> From: [email protected] [mailto:[email protected]] On Behalf Of
> Michael Barnes
> Sent: Monday, May 14, 2012 12:05 PM
> To: [email protected]
> Cc: [email protected]
> Subject: Re: [OSPF] ABR/ASBR with clear R-bit?
> 
> Hi Tanmoy,
> 
> I agree, this is an interesting use case. However, we must be careful to
> handle it correctly. Consider, for example, when the only path to the
> forwarding address is through the ASBR which has the R-bit cleared.
> Routers in the same area as the ASBR would be able to determine this
> without any trouble and not use the ASBR for transit. We need to
> consider that the ASBR may be advertising local routes as externals, and
> these should be reachable via the ASBR even when the R-bit is clear. If
> the forwarding address shares the same prefix, then it would also be
> reachable in this scenario. So how do other routers determine which
> external prefixes are reachable, or not, via the ASBR with the R-bit
> cleared? In particular, the routers which are in other areas?
> 
> I can think of a couple of ways. A simple one would be for the ASBR to
> advertise the FA with an infinite metric. This would allow routers to
> calculate another path to the FA, if one is available, while ensuring
> the the ASBR itself would not be used as the transit to the FA. While at
> the same time allowing reachability to prefixes local to the ASBR, of
> course the local prefixes would not be advertised with the same FA. 
> 
> Thanks,
> Michael
> 
> ------ Original Message ------
> Received: Mon, 14 May 2012 07:20:09 AM PDT
> From: Tanmoy <[email protected]>
> To: 'Michael Barnes' <[email protected]>, [email protected]
> Subject: RE: [OSPF] ABR/ASBR with clear R-bit?
> 
> > Hi Michael,
> > 
> > There seems to be at least 1 use case where it would be required to 
> > install the ASE/NSSA routes advertised by a router with R bit clear. 
> > If the
> ASE/NSSA
> > routes have a forwarding address, then those destinations may be 
> > reachable directly bypassing the advertising router and could prove to
> be useful.
> > 
> > Regards,
> > Tanmoy.
> > 
> > -----Original Message-----
> > From: [email protected] [mailto:[email protected]] On Behalf 
> > Of Michael Barnes
> > Sent: Thursday, May 10, 2012 8:27 PM
> > To: [email protected]
> > Subject: [OSPF] ABR/ASBR with clear R-bit?
> > 
> > Hello Folks,
> > 
> > Something which is not discussed in RFC5340 is how to treat Inter-Area
> 
> > or External advertisements from an ABR/ASBR which has the R-bit clear 
> > in its Router LSA. My initial thinking was that other routers should 
> > simply ignore those advertisements.
> > 
> > However it later occurred to me that it might be desirable to reach 
> > those destinations if they are on links directly connected to the 
> > advertising router. And if those Inter-Area or External routes will be
> 
> > installed in the routing tables of other routers, the ABR/ASBR should 
> > stop advertising prefixes which are not on its own interfaces.
> > 
> > I think this deserves some discussion and we should have consensus so 
> > that all routers behave the same way.
> > 
> > Regards,
> > Michael
> > _______________________________________________
> > OSPF mailing list
> > [email protected]
> > https://www.ietf.org/mailman/listinfo/ospf
> > 
> 
> 
> _______________________________________________
> OSPF mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/ospf


_______________________________________________
OSPF mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/ospf

Reply via email to