Mark, On Wed, 2009-09-02 at 11:36 +1000, Mark Andrews wrote: > > Feel free to do that with networks you operate. This is a huge cost, if > > you compare it to a zone file with a $RANGE statement, which is what we > > have today. > > How is it a huge cost? Please tell me. Most of the zones would > have 100 max records in them to cover a /56. You mean ISP's can't > support a 100 records / customer? You could have a policy which > says if you go over 150 records you need to run your own servers > which we will delegate to and we will warn you once you reach 100.
<snip/> > Again where is the huge cost. These are not pre-populated zones. > They get populated by the end user equipment. Windows already does > this by default. The cost is, as always: 1. Building the system to handle this new service 2. Maintaining the system for this service 3. Customer support costs (training support staff, handling customer questions and problems, and so on) Today the cost for all of these is effectively 0, because ISPs have to run DNS anyway, and your entire functionality is that I see "s5590c001.adsl.wanadoo.nl" when I ping my IP address. Adding a single line to a zone file for each customer block to support this is extremely cheap. Note that I said the cost is huge COMPARED to current practice. While the cost of putting together your magic DNS system may only be a few hundred dollars of time, this is still a lot of money compared to 0. (And likely the minimum cost would be thousands of dollars of time, even for small ISPs - the ones who can afford to spend it the least.) Again, nobody is arguing to deprecate reverse DNS for IPv6 here. Instead the idea is that it should be clear that since current BCP for IPv4 reverse doesn't work, please don't worry about it for IPv6. -- Shane _______________________________________________ DNSOP mailing list DNSOP@ietf.org https://www.ietf.org/mailman/listinfo/dnsop