Re: FYI: DNSOPS presentation

[bmanning@vacation.karoshi.com]

On Sat, Apr 03, 2010 at 07:45:44AM +1300, Brian E Carpenter wrote:
> Bill,
> 
> On 2010-04-03 05:55, bmanning@vacation.karoshi.com wrote:
> > On Fri, Apr 02, 2010 at 06:09:53PM +0200, Rimi Despris wrote:
> >>> Surely a better hack would be for recursive resolvers in IPv6-broken
> >>> networks not to serve up AAAA records at all?
> >> Agreed.
> >>
> >> RD
> > 
> > 	aren't you making the assumption that the recursive resolver is
> > 	on/in the same broadcast domain as the stub which it is serving?
> > 	how can you ensure that the IP reachability of the stub is "semantically
> > 	simialar"** to the IP reachability of the recursive resolver?
> > 
> > 	the DNS has no hooks into topological reachability. 
> 
> Correct, but a hack in this area by definition can't cover all cases.

	does it even cover a statistically significant fraction of cases?
> 
> Er, much better to mend the IPv6 brokenness, of course.

	then why not spend the energy/resouce on ensuring v6 transport
	works instead of wasting this on what is clearly a DNS haq for 
	a temporary condition?  Or is this a case of "when all we have is
	the DNS, everything looks like a query/response"?

> 
>    Brian
> 
> > 
> > ** identical, equivalent, etc... See IDN equivalence
> > 
> > --bill
> >