Re: Challenges for the BGP MIBv2

[Jeffrey Haas <jhaas@nexthop.com>]

Juergen,

On Wed, Sep 29, 2004 at 08:34:07PM +0200, Juergen Schoenwaelder wrote:
> The IF-MIB is a prominent example where the ifTable and the ifXTable are
> rooted in very different branches of the OID space. For real management
> applications, this does not matter at all. BTW, the new module will be
> rooted below mib-2 anyway according to the MIB review guidelines.

Could you clarify this please.  Do you mean that if I create a MIB
meant to extend BGP, it would be rooted at mib-2 and not rooted
at a defined extention point within BGP?

If so, why?

> I looked through the existing BGP4-MIB today (for some other reason).
> It is IPv4 specific - do you plan to make it IP version independent?
> In that case, you have to rewrite it anyway and the whole discussion
> about extensions of the existing MIB becomes a mood point.

The original version of the BGP MIB, second version, has expired (at 
draft-ietf-idr-bgp4-mibv2-05.txt if you care to go digging).  Or I 
could send it to you.  Work on it languished for over a year since
we were getting the last of the details for the first version of the
MIB straightened out for standards purposes.  Since the first version
has finally passed IESG last call, I'm focusing on the second version.

The second version not only is address family independant, it is the
first attempt to deal with the numerous extensions to BGP-4.  Unfortunately,
we get new extensions "frequently" by IETF standards and they progress
at wildly different rates on the standards track.

It also attempts to add in configuration objects for BGP.  However,
given that the implementation report for the first version of the
MIB indicated a complete lack of support for the write portion of
the two objects that permitted it, I am reconsidering this part of the
effort and will consult the WG on the matter.  The presence of
configuration objects for l3vpn has half convinced me to proceed with
BGP configuration objects.

An unfortunate fact of life is that configuration objects for BGP
are not very helpful without some way to configure routing policy.  
I don't believe that is in scope for the IDR WG.

-- 
Jeff Haas 
NextHop Technologies