Re: [Idr] WG Last Call foir draft-ietf-idr-bgp-extended-messages (11/12 to 11/26)

[Jeffrey Haas <[e-mail]>]

Robert,

On Tue, Nov 28, 2017 at 08:25:35PM +0100, Robert Raszuk wrote:
> > The original scenario for bgpsec is one where we had significant concerns
> > about too-small UPDATE messages causing issues.  We had to address the
> > issues there.  And we already know that as a feature it has incremental
> > deployment challenges.
> 
> Well if this is still for BGPSEC I think your suggestion to
> treat-as-withdraw is not
> that good. I do prefer to get unsecured reachability then no reachability
> at all.

If a bgpsec speaker chooses to "downgrade" its reachability, that's its
choice.  In such a case, it would no longer be trying to overflow the 4k
PDU.

Similarly, routers that limit AS_PATH length choosing to not propagate (and
treat as withdraw) an AS_PATH that exceeds its local limit is a local
choice.

> And if you make it treat-as-withdraw as MUST even if BGSEC itself says to
> drop
> crypto when it does not fit it is quite unlikely that anyone will even try
> to secure
> BGP UPDATES any time soon.

bgpsec implementations are likely to deploy extended messages in tandem.  I
don't find the argument you seem to be making compelling.

-- Jeff