Re: [Sidrops] WGLC - draft-ietf-sidrops-validating-bgp-speaker - ENDS 09/07/2018 - Sept 7th 2018

[Nick Hilliard <[e-mail]>]

Randy Bush wrote on 05/09/2018 07:07:
> and it is not only ov-clarify.  test whether your fave implementation
> re-evaluates a bgp prefix when a roa change comes in over rpki-rtr.  the
> messy story goes on.
[...]
> what is nice is that the ixp-provided filter does not have the same
> problems as above.

really it does.  As Job suggested, the majority of ixp route servers run 
BIRD, and taking the example you mention, one of BIRD's known 
limitations is that it does not handle revalidation.  Another limitation 
would be that it doesn't handle aggregators as the last element in the 
as path.

In general, if the IXP uses any particular bgp stack (whether ios-xe, 
junos, bird, etc), it will be stuck with the bugs and shortcomings of 
that particular implementation, and under the proposals of the 
validating-bgp-speaker draft, everyone at the ixp will be subject to 
those particular bugs and shortcomings.  It's not valid to say that the 
IXP rpki implementation will be any better than a "hardware" router 
because - as Job pointed out already - it's just software under the hood.

Nick