Re: [mif] [dnsext] bare names (was: 2nd Last Call for MIF DNS server selection document)

[Mark Andrews <[e-mail]>]

In message <[e-mail]>, Margaret Wass
erman writes:
> 
> Hi Keith,
> 
> On Oct 19, 2011, at 9:48 AM, Keith Moore wrote:
> > split-brain DNS is an abomination that should be eradicated from the planet
> .
> 
> Split DNS exists and is in wide-spread use, and that is just a fact.  We don'
> t have the power to eradicate it, nor do we currently have a better solution 
> for the types of things that people use split DNS for.
> 
> Margaret

That said there is little technical need for split brain with IPv6
only networks.  IPv4 networks and RFC 1918 addresses created a
technical problem (ambigious address use) that split brain DNS
addresses.

Publishing ULA addresses on the public internet shouldn't cause
problem.  One could even publish link local addresses on the public
internet if we added a globally unique differentiator for the link
to the records.

e.g.
	AAAA <domain> 

The domain would announced in RAs so that nodes on the link could
correctly filter the responses.  It would also allow getaddrinfo
to fill in scope.

8 byte address records would have solved the RFC 1918 address issue.
First 4 bytes are the A records and the next 4 are the public address
of the NAT or 0.0.0.0.  Resolvers would just filter out anything
that wasn't to 0.0.0.0 or their public NAT address.  One could have
even used them to route throuh the NAT using LSR.

> _______________________________________________
> dnsext mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/dnsext
-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: [phone]                 INTERNET: [e-mail]