named timeout when trying to resolve a certain (definitely exisiting) hostname on the internet

[Barry Margolin]

In article <9s9ne3$re7 at pub3.rc.vix.com>,  <Mark_Andrews at isc.org> wrote:
>
>	This is a well known limitation of BIND 4 and BIND 8 (<
>	8.3) where named doesn't chase down chains of nameservers.
>
>	To locate axa.de you have to find the addresses for
>	dns01.pironet-ndh.com, dns02.pironet-ndh.com and
>	dns03.pironet-ndh.com.  Now to find the addresses for these
>	servers you have to find the addresses for dns.ndh.net and
>	dns2.ndh.net.  You also have to know the addresses for all
>	the parents.
>
>	This configuration is known not to work with the above
>	servers (when starting from a empty cache). I've yet to
>	understand why anyone would want to configure the DNS like
>	this.  All it does is introduce extra failure points and
>	increase the amount of lookups (and hence time) it takes
>	to resolve a address.  ISP should know better.  Configuring
>	the nameservers like this is a dis-service for their
>	customers.  This is not the worst case, I've seen chains
>	5 hops long before you start going back up the heirarchy.

What alternative does the ISP have?  Should we refuse to host domains
outside the GTLD's?  Or are we supposed to have servers in every country
TLD that our customers might want us to host domains in?

-- 
Barry Margolin, barmar at genuity.net
Genuity, Woburn, MA
*** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to newsgroups.
Please DON'T copy followups to me -- I'll assume it wasn't posted to the group.