I hate to butt in here, I've been listening to these discussions for
some time. (I am incredibly impressed with how smart everyone in these
IETF groups is).
But, what about NMS directories that contain devices (non-computer),
physical
location, automations, histories, provisioning, and acquisition information?
There seems to be a debate to split "DNS" from "Directory" services,
whereas, in long term, it is inevitable that DNS will merge with
Directory services, even if current technology isn't that way.
Pushing for a conceptual split in theory will slow the convergence
of DNS/Directory. I would argue, that this convergence is very valuable,
and is the natural progression (in long term).
The concept of a directory should be a large database, with pre-defined
standards for how different types of common knowledge is built in,
but should also allow for user defined types, for which no existing
standard exists. DNS should ultimately become one standard data type
of this theoretical global directory. As should what we want to do,
which is storing automations, configurations, and histories in the
directory.
The IETF community should be aware that it is probable that it is
impossible to predict *what* one would want to store in a Directory,
but that there is standard information that should be well-defined
to extract from the directory.
This is a passionate issue for us as no existing directory implementations
have supported all of the requirements for our NMS and we built on SQL
databases.
Ultimately, we believe a directory service should be just like a massive
SQL database, but include standard "Tables" for standard things, like
user accounts, DNS, computers, etc. It should all be centrally accessible
in a common manner, but support custom user types, in addition to standards
for standard things.
It is dangerous to say "DNS is not Directory". While in today's existing
implementations that is true, ask yourself the question, in the long
term, will this still be true? And if it will be, is that good or bad?
I would argue, that in the long term, DNS *should* merge with Directory
services.
Kyle Lussier
Directory service = a software system that responds to requests
for information about entities, e.g. people in an organization.
It's a system for managing access to computer resources, keeping
track of the users of a network,... from a single point of
administration. It allows a network administrator to set up and
control a database of users and resources and manage them using a
directory (by example with an easy-to-use GUI, Graphical User
Interface). Users, computers, sites,... can be added, updated and
managed centrally ; applications can be distributed electronically.
Microsoft Active Directory, Network Information Service (NIS),
Novell Directory Service (NDS) and X.500 are examples of
directory services.
--
Address registry = a registry of numbers or addresses with some
corresponding data, e.g. names. Such a registry helps maintain
names, which are identifiers that are mapped to numbers or addresses.
Let's say a Directory Service is multi-dimensional, in the sense
it involves many types of data, many levels of information you
have to search in, while an Address Registry has one dimension,
in the sense it just maps addresses to their corresponding name,
like a telephone registry, DNS, WINS, the "hosts" file or the
"lmhosts" file.
--
So, what is DNS? In the TCP/IP world, the Domain Name System
(DNS) is a distributed database that provides the mapping between
IP addresses and hostnames. It's just an address registry.
--