Harald,

> Harald Tveit Alvestrand wrote:
> [Dual-headed DNS]
> That is well known.
> It's also a pain to configure,

No argument here; and still lots of people are doing it.


> My question to you is whether:
> - the use of site-local FORCES you to use split DNS, even
> if you otherwise don't need to

Not at all:
- If the host has a site-local only address (which I recommend, although
this is not consensual) all is needed is the existing one-headed DNS, no
change.
- If the host has both a site-local and a global address, this issue has
nothing to do with site-locals and everything to do with multihoming /
multiaddressing the host. In other words, the need for dual-headed DNS
and the issues pertaining to source and destination address selection
would be the same if the host had two global addresses instead of a
site-local and a global. This is a network design issue.


> - the use of site-local and split-DNS FORCES you to let the
> boundaries of the site follow the boundaries of your security
> perimeter, or suffer the N*2 problem of having to manage four
> categories of names rather than two

Probably. It appears though that postings by different people seem to
converge about the idea that the boundaries of the site (as in
site-local) match the administrative boundaries of the organization.
There is an issue with semantics but not with the concept itself.


> (btw, IMNSHO, the security argument for split DNS is security
> through obscurity - it only protects you against the stupid
> bad guys....)

True enough, although it is better to protect against the stupid bad
guys than not to; and the wide spread of IPv4 dual-headed DNS today that
has nothing to do with site-locals is likely a good prediction of the
spread of IPv6 dual-headed DNS for the same reason.

This is the same issue with site-locals themselves: A little reality
check shows that they will be used no matter what.

Which brings me to the point I was trying to make in the first place:
People that will use site-locals for the
"security-by-not-routable-addresses" will already have dual-headed DNS
for the "security-by-obscurity", no extra work here.

Michel.


--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to