[DBWG] RIPE proposed changes to the routing registry

[Daniel Shaw]

On 17/05/2018, 12:19, Job Snijders	typed:
> 
>> 
>> The decision is not about technical limitations, but rather more about
>> what is the current understood business rules and processes.
> 
> This answer surprises me, because earlier in the thread you state:
> 
>    "It is not impossible to create these objects, but it does involve
>    manual intervention by AFRINIC staff"
> 
> so at first glance it seems this is not about business rules, but
> automating an operation that is currently performed manually. AfriNIC
> already allows route-objects in the AFRINIC IRR where the ASN in the
> "origin:" is a non-AfriNIC ASN.
> 

Ah. Let me explain :)

So, yes, while the end result is allowed, the current process is that the manual step by the person is some check or verification that the while the ASN may be non-AFRINIC, it is still and ASN that is allocated by some RIR and at that RIR has some association with/to/from the person or organisation that has requested the route(6) object. In other words some form of manual authentication.

So yes, AFRINIC allows (some) route-objects in the AFRINIC IRR where the ASN in the "origin:" is a non-AFRINIC ASN. But not *any* non-AFRINIC ASN, or even any ASN.

Of course, the necessity of this at all is debatable. (As you point out on the dbwg at afrinic.net list).

However, given that a check like this *is* being done today, so far it's been felt easier to have a human do it, than to automate around all the edge cases. Like for example where the "foreign" ASN is registered to a subsidiary or parent company, so there is no *direct* one-to-one mapping of Org name in WHOIS, and knowledge of multi-national business relationships may be needed to determine the link.

> 
> I believe the check causes more work for AfriNIC staff and the end users
> than necessary (or will in the very near future when RIPE closes the
> loophole). The check actually may _cause_ more communication noise
> because over and over again the AfriNIC staff will have to explain that
> the webinterface can't accept the route-objects and manual intervention
> is needed.

Thanks! Very good to have this clearly in text to pass on.

Cheers,
Daniel