[DBWG] RIPE proposed changes to the routing registry

[Job Snijders]

On Thu, May 17, 2018 at 10:13:22AM +0400, Daniel Shaw wrote:
> In fact, my understanding is that is actually how the RIPE DB will
> work in future.
> 
> This of course means that any resource holder could then create IRR
> objects that state their resources are ok to originate from any ASN at
> all. So the question is then for ASN holders: How does a given ASN
> feel about route(6) objects existing with their ASN as origin that
> they have nothing to do with or any control over?
> 
> This is an honest question, and not being the operator of a large ASN
> I truly do not know the answer.
> 
> Personally I see little risk in this, but I'd like to hear from large
> networks, their opinion(s).

I work for a fairly large network.

There is no issue, and there are TONS of databases that do not
check the ASN in the "origin:" attribute. So RIPE is hardly the first to
drop this requirement.

The origin-check does NOT exist in at least the following IRR databases:
ALTDB, APNIC, ARIN IRR, ARIN WHOIS, BBOI, BELL, LEVEL3, RADB, RGNET,
NTTCOM, and TC (probably more DBs too).

A second thought is that in the RPKI, when an end user produces a ROA -
they do not need permission or involvement from the ASN owner. In a RPKI
ROA the prefix owner can permit any ASN as the Origin ASN! With IRR
route objects we should not complicate matters.

Kind regards,

Job