RE: Wide-Open MADRID

At 11:26 -0700 6/8/04, Greg Connor wrote:
> I think I was thinking about AS in order to avoid going to Whois... if the
> block info could be extracted from DNS that would be better for performance
> reasons.  Perhaps looking up the first and last IP in in-addr.arpa to see if
> it is delegated to the same place?  Or is there a better way to learn the size
> of the ARIN block using only DNS queries?
Well, AS's aren't in DNS - there's no as.arpa.  (It's been suggested 
before, and it comes up now and then, but it then goes away.  I don't 
have the story why.)
When it comes to the reverse map of DNS, DNS is an authority on what 
is present but not a strong authority on what is what.  The 
reliability of the data that is present there is as reliable as the 
forward space - it is the missing pieces of the reverse map that are 
a problem.  Getting address space from the RIR's does not require 
that you have reverse map name servers.
In general, DNS is a poor tool for IP address management at any 
scale.  (I say this not from the perspective of an RIR staff member.) 
DNS only allows IPv4 to be "easily" subnetted at /8, /16/, and /24, 
the "hard" option showing finer granularity is much, umm, harder and 
therefore rare.  In v6, it's a bit easier to subnet closer to 
"reality" but the very thought of v6 reverse map is still being 
debated.  And that's just when you are trying to represent subnets in 
DNS.
The only good way to get block sizes from ARIN is via the whois.  (In 
the future, it'd be CRISP WG's protocol.)  The rationale is that 
whois is a better listing of what's in a registry's database than 
DNS.  The DNS only represents a report of the database limited to 
zone delegations and name servers.  whois is a more comprehensive 
report.  E.g., a domain "on hold" won't be in DNS but may be seen in 
whois.  (The latter point came up in the PROVREG WG when it was 
generating EPP.)
One other factoid - ARIN talks of address ranges, not CIDR blocks, 
for a reason.  There are some ranges that are let's say, quirky, 
because of typos (345.34.13.1-345.34.13.255, omitting a ".0" at the 
start) or because of unusual circumstances.
Another way to think of this is that in the forward map, a 
registrant's goal is to have a name in DNS, in the reverse map, a 
registrant's goal is to have an entry in the route tables.  The only 
commonality between the two is what is in the whois. (Or CRISP.)
--
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Edward Lewis                                            +1-703-227-9854
ARIN Research Engineer

Even the voices inside my head are refusing to talk to me anymore.