Re: [IETFMIBS] Extending InetAddress(Type) for BGP Multicast VPNs

[Juergen Schoenwaelder <[e-mail]>]

On Tue, Jan 10, 2012 at 06:28:09AM -0800, Jeff Haas wrote:
> Brian,
> 
> Since your response is closest to my intentions, I'll answer here:
> 
> On Jan 10, 2012, at 7:58 AM, Brian Haberman wrote:
> > I suspect that Jeff wants to introduce 7 new values for InetAddressType
> > that allows him to represent these seven types of routes:
> > 
> >     + 1 - Intra-AS I-PMSI A-D route;
> >     + 2 - Inter-AS I-PMSI A-D route;
> >     + 3 - S-PMSI A-D route;
> >     + 4 - Leaf A-D route;
> >     + 5 - Source Active A-D route.
> >     + 6 - Shared Tree Join route;
> >     + 7 - Source Tree Join route;
> > 
> > I would have to see a more concrete proposal, but it seems a unwieldy
> > given that each of the above types could occur for both IPv4 and IPv6
> > addresses.  So that would require more than seven new values for
> > InetAddressType.  Unless I am not understanding the referenced draft
> > correctly…
> 
> You and Juergen are understanding this perfectly fine.
> 
> Also, as I expected, you're all not very fond of the idea (using differing wording in the various responses, but it's clear).  That saved having the discussion complicated by providing sample SMI.
> 
> What this means is, as the editor of the BGP MIB, I now have to ask and answer a slightly different problem.  I made use of the InetAddress MIB so that we could take advantage of the flexibility to cover both IPv4 and IPv6 reachability for our peering sessions as well as our prefixes.  However, BGP is capable of carrying more reachability than simply IPv4 or IPv6; in the example case, it's MVPN.  We also support things like VPLS.
> 
> This seems to suggest that BGP needs to copy and fork InetAddressType/InetAddress (and perhaps the prefix length TC as well) for a BGP specific MIB that can accommodate non-IP prefixes.
> 
> I would love an alternate suggestion to doing this since I do not relish the idea of maintaining such a TC MIB indefinitely. :-)
> 
> (I also understood that was exactly what I was asking the owners of INET-ADDRESS-MIB to do.)
> 
> One property such a BGP-specific MIB should have is to use the same code points as the INET-ADDRESS-MIB where possible.  This would enable code re-use.
> 
> Is my logic correct?

If the reachability info BGP is carrying turns out to be BGP specific,
then having a BGP specific TC to represent it seems quite OK. Is any
non-BGP protocol using the formats and the encodings I looked at
earlier this morning? If not, these formats seems to be rather BGP
specific.

/js

-- 
Juergen Schoenwaelder           Jacobs University Bremen gGmbH
Phone: [phone]         Campus Ring 1, 28759 Bremen, Germany
Fax:   [phone]         <http://www.jacobs-university.de/>