[arin-ppml] Fee structures for ARIN

[Owen DeLong]

On Oct 30, 2011, at 12:22 PM, William Herrin wrote:

> On Sun, Oct 30, 2011 at 1:23 PM, Owen DeLong <owen at delong.com> wrote:
>> Sorry, I'm having trouble understanding here how an NS record covering
>> a /32 is going to consume more than an NS record covering a /48.
>> 
>> Can you please explain that one to me?
> 
> Which record would say is requested more frequently?
> 
> com NS or delong.com NS?

Depends on the server. My nameservers answer many more queries for
DELONG.COM. I'm guessing that the GTLD-SERVERS also answer more
queries for DELONG.COM (since they really shouldn't see queries for
COM NS most of the time). OTOH, the ROOT-SERVERS probably see
far more queries for COM and almost none for DELONG.COM.

> 3.2.192.in-addr.arpa NS or 192.in-addr.arpa NS?
> 

From ARIN's perspective, likely 3.2.192.in-addr.arpa. NS

192.in-addr.arpa is probably cached very quickly by the vast majority of
resolvers and likely ARIN sees some, but, not very many queries for this
when compared to the queries for delegated prefixes.

In a more realistic perspective, do you really believe that, for example,
the queries for 192.128.0.0/16 would somehow be significantly more load
than the queries for 192.129.128.0/17?

I would tend to bet that the queries for 192.128.0.0/16 would quite likely
be less, since the former requires only a single NS record set while
the latter requires ARIN to hold 128 NS delegation record sets each
of which must be cached individually by the resolving hosts.

In other words, yes, there is possibly some relationship to the prefix size
in terms of the DNS load, but, it is most definitely not a linear relationship
and more than likely not directly related even on a logarithmic or other
basis.

Owen