Hi,

> The problem in such discussions is "harm to whom?"  The harm is that
> is we incorrectly solve this, we harm the overall architecture of the
> Internet.  The issues I read pertained to the host, not the DNS,
> hence, to prevent harm to the architecture, the solution lies in the
> host.

I see this rather so that a network causing issues for the host ought to 
provide assistance for hosts to overcome those issues.

> There are a number of problems with the question.  One, split-DNS
> isn't defined and that hinders talking about policies about its use.
> Two, I don't understand "DNS server selection" as something that one
> does.  Three, split-DNS is a thing that usually is only introduced
> where it is a solution to problems, if it was causing problems it
> would be removed.

>From 10000 feet view some DNS servers have information others do not. Isn't 
>that enough of high level definition of split-DNS? 

The DNS server selection is a mandatory feature for every host that receives 
more than 1 DNS server address from any sources. In simplest case it just means 
overwriting /etc/resolv.conf with the latest info received from anywhere.. 

There was a good email "two reasons to have split-DNS" by Mark Andrews last 
Friday. Won't those reasons remain valid reasons even in the presence of 
multihomed hosts, hence unlikely to disappear any time soon? Hence need to find 
a patch to the problem the solution caused. 

> And that is why there's no clear answer.  There's no agreed
> definition over split-DNS.

Right, but doesn't split-DNS always mean some DNS servers have some information 
others do not?

What other criteria of selecting a DNS server could be in addition to FQDN and 
IP-addresses (ranges, subnets)?

Best regards,

Teemu
_______________________________________________
DNSOP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dnsop

Reply via email to