[DNSOP] Late to the "special use labels" party

[william manning <[e-mail]>]

actually, these ideas touch on a few threads that seem to (still) be flying
around.  I expect to turn this into an ID, headed for Informational -
possibly to the ISE.

comments and constructive input appreciated.
--------
/Wm

What is the Domain Name System (DNS)?

The DNS was created to provide a scalable system for providing a mapping
between the name of an instance and the location or address of that
instance.  It has three essential elements; an ephemeral namespace, servers
which instanciate the namespace;  a suite of protocols that allow a client
or resolver to ask the servers questions about the namespace.   All three
of these are required to say that the system is or is part of  DNS.  A
fourth presumption is that there is always on, always connected
reachability across the DNS namespace.

The suite of protocols used between resolvers and servers, as well as
server and cache behaviour are within the perview of the IETF and its
working groups.

The Namespace is designed as an inverted tree, with a single root context
per protocol.  Although other protocols were invisioned at the outset, to
day they are primarily vestigial, at least as far as the IETF is
concerned.  There is a single root, and one namespace for the DNS, as far
as the IETF is concerned.

Traditionally, the IETF did not concern itself with the contents of the
namespace, leaving the management of the delegation points to the zone
maintainers, since this was always going to be a matter of local
preference.

These constructs, in unison, have created the global DNS as we know it.
However, the tools are so useful, others have borrowed from them for other
work.

What is NOT (strictly) the DNS.

It is possible, and has been implemented for decades, to change out parts
of the DNS namespace for ones own version.  Split-DNS enables DNS-like
services for private spaces not connected to the Internet.  Often these
private namespaces augment the Internet namespace with other, non-Internet
names.  As far as the servers and resolvers are concerned, they still use
the default DNS protocols.  It is hard to tell if one is or is not using
the DNS or a faximile just from the resolver side.

Others want to use the DNS namespace, but invent their own protocols for
server/resolver communication.   Some want to change out the concept of
servers/resolvers, but use the namespace.

NONE of these hybrids is DNS.  They are DNS-like, some are parasitic some
are symbiotic, but they are not DNS.

It is a mistake for the IETF to treat these non-DNS issues as DNS related
and it is a mistake for the IETF to get involved in dictating to zone
maintainers what labels they may or may not chose to  put into their
delegations.  As long as the communications protocols are ok with the
labels.

Those involved with the DNS should also avoid mission creep into how other
applications may or may not chose to utilize the names/labels returned from
a DNS query.  If the DNS working groups stay focused on staying within
their remit, other application developers will not have to be so concerned
with what the DNS does or does not, and if they have to develop their own
systems.

--------------------------------------------------------------------------------------