Re: [Idr] draft-jakma-mrai-dep-00 (fwd)

Hi Pedro,

Your assessment is correct. 

However, within a particular network (say, providing BGP/VPN service),
the reachability and unreachability may not translate to a single event.
In the below simplified topology, CE1 is dual-homed to PE1 and PE2
(having the VRFs with unique RDs) -

      CE1-----PE1------------PE3-----CE
	 \------PE2------------/

CE1-PE1 link failure results in withdrawn of bunch of routes from PE1 to
PE3, which needs to receive this event as quickly as possible (related
to any other previous advertisements PE1 may have sent to PE3). 

This stays as a single event, even if there are RR in the network (given
the unique RD per VRF per PE).

Needless to say that the above scenario becomes trivial with MRAI = 0
for iBGP.

Cheers,
Rajiv

> -----Original Message-----
> From: idr-bounces@ietf.org [mailto:idr-bounces@ietf.org] On Behalf Of
> Pedro Marques
> Sent: Wednesday, November 26, 2008 1:08 AM
> To: Jeffrey Haas
> Cc: Yakov Rekhter; idr@ietf.org
> Subject: Re: [Idr] draft-jakma-mrai-dep-00 (fwd)
> 
> 
> On Nov 25, 2008, at 8:20 PM, Jeffrey Haas wrote:
> 
> > Yakov,
> >
> > On Tue, Nov 25, 2008 at 05:40:33PM -0800, Yakov Rekhter wrote:
> >> With my WG chair hat off, I think the document more or less
> satisfies
> >> the requirements. The only two change I would suggest to make are
> >> (a) to
> >> eliminate a special case for Withdrawn Routes (this is based on
> >> some discussion that happened few years ago initiated by Pedro
> >> Marques),
> >
> > By this, do you mean you don't believe that MRAI should be
separately
> > configurable for reachability and unreachability?
> 
> The rational for not treating them differently is that a single event
> as it propagates through the network gets expressed as both
> reachability and withdrawal information.
> 
> The typical example being an SP that acquires a newly preferred peer
> route which them causes it to send withdrawals to other peers
> (reachable converts to unreachable).
> 
> My understanding is that this behavior has been well document in
> practical observations of BGP convergence.
> 
> In general my recommendation for MRAI values would be 0 for eBGP and 0
> for iBGP. As the draft correctly explains calculating a value that
> best fits the network is a very tricky business.
> 
> regards,
>    Pedro.
> _______________________________________________
> Idr mailing list
> Idr@ietf.org
> https://www.ietf.org/mailman/listinfo/idr
_______________________________________________
Idr mailing list
Idr@ietf.org
https://www.ietf.org/mailman/listinfo/idr