[Idr] Re: Roman Danyliw's Discuss on draft-ietf-idr-vpn-prefix-orf-37: (with DISCUSS and COMMENT)

[Jeffrey Haas <[e-mail]>]

Robert,

> On Apr 29, 2026, at 14:40, Robert Raszuk <[e-mail]> wrote:
> 
> IMO the term "backpressure mechanism" really has a meaning in the cases where a consumer can signal the producer(s) inability to handle the delivered load. 
> 
> Here we clearly are not dealing with such situations. 
> 
> Just by pushing filtering on the information distributor (Route Reflector) asking it to drop the updates has in my opinion not much to do with real backpressure. 

If the garden hose simply sprays its water as a leak from the nozzle because it is blocked rather than draining the line, it is still providing relief from the pressure.  

> While we are at the final stage of the reviews of this experimental draft I would also like to highlight that any policy drops on the IBGP side can cause loops. It is vastly different from dropping VPN updates due to non-intersecting RTs irrespective if done locally inbound on PE or remotely signalled by RTC. 

It is, however, undistinguishable from the receiving VRF on a PE from discarding the reachability itself as part of local procedures.

If you believe there are situations providing for loops, perhaps requiring RFC 6368 procedures which are lightly deployed, the IESG might find an example helpful.  Especially in contrast to drops due to local policy.

-- Jeff