🚀 go-pugleaf

In article <1993Jan7.022756.25843@ssigv.UUCP> gdonl@sunrise.ssi1.com (Don Lewis) writes:
>In article <BjQPrA-FBh107h@boombox.apana.org.au> David.Keegel@apana.org.au writes:
>>it be nice if BIND cached negative information like "UUCP. -- unknown
>>domain".)
>
>Even if BIND had negative caching, it wouldn't help much in this case.
>The problem is that there wouldn't be any queries for just "UUCP".  Your
>local name server would forward queries for foo.UUCP and bar.UUCP to
>the root servers, which would return a no such domain response.  Your
>local server would cache this information for foo.UUCP and bar.UUCP,
>but when it received a query for baz.UUCP, it would still forward it
>to the root servers.

Not so.  Assume BIND does negative caching.

1. resolver gets query for "foo.UUCP".
2. Since it does not know this data, it first must find it out from
   the root nameservers since this is not in same top-level domain.
3. The nameserver contacts a root nameserver to find out who "foo.UUCP"
   is.  In the NXDOMAIN response is "how far" the root nameserver got
   down the tree before it hit the invalid domain.  (For example if you
   query "foo.UUCP" you get ".".  If you query "bogus.us" you get
   ".US".  For "bogus.pa.us" you get "pa.us") Thus the local nameserver
   knows that there is not a ".UUCP" domain, and can therefore cache that
   information.

In the answer is also the TTL value.  It would be safe for the local
nameserver to use this TTL value for the negative cache entry.

(now all that's left is to implement it.  :-) )

[followups to comp.protocols.tcp-ip.domains.

--Dave
-- 
System Administrator, Population Research Institute    barr@pop.psu.edu
What does it profit a man if he gains the whole world, but still runs DOS?