IPv6 blocks for micro-allocation

Nick Hilliard nick-lists at netability.ie
Wed Jun 4 01:33:36 CEST 2008

> Yes, and to the unwashed little operator like me, it is completely
> beyond me, why there isn't a hierarchic authentication and authorization
> scheme, with IANA holding the toplevel "maintainer" objects for
> inet(6)nums and route(6) objects (and others).
> Almost all RIRs have adopted the RIPE DB software, only(?) ARIN still
> using some vintage system, not tying IRR data to IP space registration
> data.
> Must be some kind of Not Invented Here syndrome...
> Or I must be overlooking something...

it's a bit more involved than that.  ARIN inherited some revision of the 
InterNIC database, poked it to the point of complete incompatibility with 
anything whatever, and currently serve data in a format which is rather 
less than compatible with anything at all.  It was a great pity that they 
didn't take the chance to migrate to either the Merit or the RIPE whois 
database daemon at the InterNIC->ARIN split.  APNIC and Afrinic use fairly 
recent RIPE daemons, and LACNIC front-ended everything with Joint Whois, 
which means that you could get anything at all from them.

There are lots of other routing registries out there.  Jos Boumans gave an 
interesting talk at RIPE-56 about the difficulties faced in mirroring the 
systems, but the main problem appears to be that there are too many of them 
around, supporting different features, and often with conflicting data for 
the same lookup keys.  He mentioned a figure of 40+.

Also, while lots of these IRRs use the RIPE code, lots more use the IRRd 
code.  And many use slightly older code versions which don't support later 
features.  For example, the Level 3 IRR (rr.level3.net) uses RIPEdb 
3.0.0a13, which dates from the pre-RPSL days.  It's pretty ancient code at 
this stage.

IRRd and RIPE support different syntax too, and it's annoying.  RADB 
natively supports server-side as-set expansion, but RIPE doesn't. RIPE 
supports a well structured mnt-routes and mnt-lower hierarchy, but RADB 
doesn't.  And so forth.  The list of grievances and incompatibilities is long.

Anyway, to deal with your question, IANA has no relationship with most of 
these routing registries.  RIPE happens to be particularly well organised, 
because it manages address ownership with the IRRDB in the same database, 
and has a relationship with IANA, so it can do funky stuff like allowing 
its users to have some level of security when dealing with routing info. 
But this can't really happen with any of the other non-RIR IRRDBs because 
they are just organisations on the net who happen to run IRRDBs; they have 
no particular relationship with IANA or ARIN in this respect.  It might be 
nice if ARIN were to dump their fossilized format and go down the route of 
all the other RIRs, but that doesn't appear likely to happen any time soon. 
  Mean-time, you can expect the current situation to stay much them same.

In fact the only thing you can say for sure about all these whois servers 
is that they listen on port 43/tcp and may provide useful information if 
given some arguments.  Beyond that, you're on your own.  Of course, getting 
people to change requires dealing with politics:  "why are you changing 
this?  it works fine".  Just like the imperial measurement system*.  It 
works fine.

Summary: it's a mess.


*incidentally, a year or two ago, someone finally gave me one good feature 
of dealing with inches and feet.  You can trisect a foot with a standard 
imperial ruler, which is something you can't do with a metric ruler.  But 
other than that, the system is utterly bogus - in my extraordinarily humble 
opinion.  Now, what were we talking about again?

More information about the ipv6-ops mailing list