Towards a Disjoint IRR

In message <9505011354.AA29144 at ncc.ripe.net>, Daniel Karrenberg writes:
> 
>   > Curtis Villamizar <curtis at ans.net> writes:
>   > 
>   > Why doesn't this seem like a problem to me?  The RADB concentrates
>   > data from CANET, MCI, ANS, anyone else with a registry in North
>   > America and also bring in data from the other continents.  
> 
> Well does it?
> Not in my view of the world.
[ ... tutorial deleted ... ]

I agree with your Q1 and Q2.

> Q3) Should RR operators register objects in their RR on behalf of users 
> without the user's  knowledge? It does not matter whether this is by
> copying from another RR or by other means.
> 
> The answer is NO because
> 	
> 	- it creates inconsistency problems
> 	- it confuses the users

There is support for secondaries, though I have't tried it, it looks
simple enough.  I agree that the US RADB should not claim to be
authoritative over RIPE for European networks, but I see no reason why
they (or ANS, or MCI) can't mirror RIPE information in a separate
database that is periodically fetched in its entirety from RIPE.

I thought I clearly mentioned a distribution of primary authority with
RADB mirroring some registries and serving as primary for providers
who don't run their own registry, or prefer to mirror the RADB (and
leave them with the registration headaches).

There is already duplication, and the IRR registries are just going to
have to resolve this over time.  Once AS690 advisories are gone
(hopefully soon), I see no reason not to drop PRDB machine generated
route object registrations in favor of a registration entered by the
end user.  We will have to provide advance notification, particularly
if a change in the recorded origin will change routing.

Long term, I see ANS as being the primary repository for ANS inet-rtr
objects, ANS aut-num, and route objects for ANS direct attached
customers.  I see the RADB as primary for customers of Sprint, PSI,
CIX, etc or any other US provider that does not run a registry.  Does
this fit nicely into your view of the world?

> Daniel

Curtis



-------- Logged at  Mon May 1 20:23:14 MET DST 1995  ---------