Christian Huitema writes ("RE: Last Call: 'Linklocal Multicast Name Resolution
(LLMNR)' to Proposed Standard"):
A key technical difference between LLMNR and the initial MDNS proposal
is precisely that LLMNR has no concept of a ".local" top level domain.
While that is true, the next statement:
Usage of LLMNR does not promote queries to this zone.
is not. LLMNR _requires_ the use of names which do not resolve in the
global DNS, and does a global DNS query for each lookup. Although
LLMNR doesn't suggest the use of `.local', given that many people
don't have a suitable domain to use, they'll use `.local' or something
else.
This is indeed a key difference between LLMNR and MDNS. MDNS assumes
that there is a special zone for local names, which would be linked to
the topology. LLMNR assumes that names are independent of the topology,
that a host called "foo.example.net" retains the same name as it move to
different locations. There were ample debates of this point in the
working group, and the decisions to "not creating special names" and
"not linking names to topology" do reflect WG consensus.
Even if that is so, this posistion doesn't seem to reflect IETF
consensus.
It is nonsense to say that a `link-local multicast name resolution
protocol' provides DNS data which is not linked to the topology !
To say that the _names_ are not linked to the topology is to miss the
point and thus to mislead.
Given that the _answers_ depend on the topology, wouldn't it be a good
idea to be able to indicate that this topology-dependent behaviour is
- or is not - desired ?
LLMNR only provides that switch - turning on and off topology-
dependent behaviour - on a per-host basis. mDNS provides it on a
per-lookup basis in a nice easy-to-configure way: specify names in
.local and get topology-dependent answers; specify other names and get
consistent answers.
Ian.
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf