On Tue, 19 Apr 2011, Marc Lampo wrote:

The present document states, "neutrally", that double-sign rollover
involves more signatures and hence, larger zone files.  While this is
true, well maintained (public) zone files should be small anyway (except
for tld's).
The average company does not offer hundreds of services on hundreds of IP
addresses.

In almost all enterprises and government locations I have seen, there is one
very large dhcp based zone, with thousands or tens of thousands of entries
in a single zone. Picking the latest customer I was at, these are the top
sizes:

[paul@bofh]$ for zone in intranet/*ldns; do wc -l $zone | sed "s/ .*$//"; done 
| sort -n -r | head -10
144711
132489
52430
41155
20312
17326
15087
11196
10289
8587

These zones, all hosted on the same primary server, would see an additional 
half a million
RRSIGs with a double-sign rollover. And for redundancy, this customer also 
loads all their
auth zones on most of their name servers globally, and it would surely mean 
they need to
upgrade hardware just to comply with this recommendation.

-1 for this recommendation. TLDs are not that special in size.

Whether this is "well maintained" or "public" I will not judge on. It is just 
what I
encounter in the wild.

A DNS administrator who leaves data like "test-mx" in hers or his public
data should learn not to publish anything not meant to be distributed.

I don't think 4641bis is supposed to make any value judgement about DNS data. 
Nor in fact,
do we really know what other new RRTYPE records (like HASTLS and TLSA) will end 
up in
the "well maintained public zone files".

4641bis should be about what is safe for the general use. If anything, TLDs are 
less
important for 4641bis because I expect TLDs to have DNS experts that can make 
their
own judgement and interpretation, whereas 4641bis will hopefully be a "default" 
setting
for most DNS servers/signers for organisations with less embedded DNS skills.

Paul
_______________________________________________
DNSOP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dnsop

Reply via email to