RE: draft-ietf-idr-bgp4-mibv2-03.txt

["Natale, Jonathan" <[e-mail]>]

thanks, jeff

i think vendor defined text would suffice and be the most extensible


> -----Original Message-----
> From: Jeffrey Haas [mailto:[e-mail]] 
> Sent: Tuesday, November 19, 2002 12:09 PM
> To: Natale, Jonathan
> Cc: [e-mail]
> Subject: Re: draft-ietf-idr-bgp4-mibv2-03.txt
> 
> 
> Jonathon,
> 
> On Tue, Nov 19, 2002 at 10:19:24AM -0500, Natale, Jonathan wrote:
> > I think you had suggested an object to indicate which 
> step/attribute was
> > used to select/not select the route as best.  I think this 
> is a good idea.
> 
> I think operators would find this particularly useful. :-)
> 
> > Has this been formally drafted anywhere?
> 
> Not at the moment.  However, the recent draft:
>     BGP Custom Decision Process 
> (draft-retana-bgp-custom-decision-00.txt)o
> 
> gives a start to this by enumerating route selection steps.
> 
> The biggest obstacle to doing this is formally enumerating the
> route selection steps and permitting them to vary from vendor to
> vendor since no one seems to be interested in getting things to the
> level of 100% consistancy.
> 
> Then again, something kinky (at least in the MIB) could be done
> such that the object contains an index into a table that contains
> the vendor's implementation of route selection.  A dump of that
> table would show what all the steps are.
> 
> Given that vendors allow tweaking of the process via knob in many
> cases, even that is probably not 100% feasible. 
> 
> The best we could probably hope for is a text description of where
> we stopped.
> 
> > I did not see it in
> > draft-ietf-idr-bgp4-mibv2-03.txt, but I may have missed it.
> 
> At this point, I'm not arbitrarily putting stuff in the v2MIB without
> group consensus.  However, given some of the recommendations made at
> the IDR session (e.g. injecting multiple route instances), the
> MIB may require serious tweaking.
> 
> -- 
> Jeff Haas 
> NextHop Technologies
>