Re: [apps-discuss] [taugh.com-standards] Comments on draft-levine-orgboundary

["Murray S. Kucherawy" <[e-mail]>]

On Sun, Jul 21, 2013 at 3:26 PM, John R Levine <[e-mail]> wrote:

> This argues for going back to the wildcard hack, with records like
>>>
>>>  *.ontario._ob.ca TXT "ontario.ca"
>>>  *.toronto.ontario._ob.ca TXT "toronto.ontario.ca"
>>>
>>> These have lousy DNS cache behavior, but the closest encloser rule means
>>> that you can find the cut point for any domain with one UDP lookup.
>>>
>>
>> Is there any hope that we could encourage implementers to cache these in
>> the application, since they're likely to change only rarely anyway?
>>
>
> The lousy cache behavior is that every different name is a different
> lookup, even if it comes from the same wildcard, e.g., foo.com,
> www.foo.com, bar.com, www.bar.com, etc.  Repeated queries for the same
> name cache just fine, give or take the cache being too small.
>
> The question is whether the rate at which hosts look for these things will
> be high enough to matter.  It seems unlikely they'd be as frequent as rDNS
> queries, and those cache just as badly.
>

Couldn't application level caching with some knowledge of this system
reduce the lousy caching?  For example, if through this system you can
learn that .com is a point below which all domains are in their own OB, you
don't need to look up _ob.com anymore (or at least not for a really long
time).

-MSK