Nebula 1994 June/NEBULA_SE.ISO/Documents/FAQ/Internet-faq/Domain.Info

Text File | 1993-01-27 | 6.9 KB | 134 lines

[Author's note: This file was originally composed as a message to
explain to a number of GEnie folks what all "that funny stuff" was
in messages that they'd seen posted on GEnie that originally came
from Usenet. Some of this information has changed since it was
originally posted, but I believe that, for the most part, the
description is still correct.
--David Kozinn  GEnie: D.KOZINN
                 uucp: uunet!vmp!monymsys!david
               domain: monymsys![e-mail] or [e-mail] ]

Usenet addresses can get very complicated. That's OK though, since
if someone tells you an address to mail to, they'll either tell you
one of two things (or a combination of them): A uucp-type path to
their machine, or a domain address. A uucp path to your machine
basically consists of "directions" to your machine from a "well-
known" address. For instance, to apply the equivalent to the US
Postal Service, in order to get mail delivered to me in Emerson
where I live, you'd probably have to write out my address by saying
something like:

        ...NYC!Hackensack!Emerson!Kozinn

Which says: Send this letter to NYC (which is a pretty well-known
place), they'll forward it to Hackensack (which is less well-known),
and they'll forward it to Emerson. The folks in Emerson will deliver
it to me. The people in NYC probably don't know where Emerson is,
but if, for example, you lived in Paramus (the next town over), you
could probably send to just Emerson!Kozinn. (This example isn't
perfect, but I think it's pretty close.) Obviously, this can get a
bit obscure if you live more than one or two "hops" off a major
location. It's also a pain.

To make things easier, and to be able to deal with things like inter-
network mail (which is sort of like UPS forwarding USPS mail!),
something called the "domain" was invented. My understanding of this
is not as good as with paths, but I'll give it a shot anyway.

Basically, a domain is some sort of organization that you belong to,
and a domain can have a number of subdomains. The advantage of
domains over paths is that they are more of a logical representation
than a physical one. To go back to my home analogy, let's imagine
that Emerson is an officially registered domain, and that I reside
within that domain. (There are procedures to get yourself
registered.) My address then would simply be:

        kozinn@Emerson

The actual syntax is a bit different when you're using unix mail,
since, for example, there are several very large domains that have
many sub-domains. For instance, in keeping with the analogy, let's
say that someone decided that there is a good chance that there's
another Emerson in the US (or the world), so in order to minimize
the problems they create a domain for each state, with each town and
city within that domain being a subdomain. Thus, my address might
read:
        [e-mail]

In "real-life", the major domain are things like .COM (for
commercial sites), .EDU (for educational sites), and a few others.

Ok fine, but how does the guy in the Smalltown, England post office
know how to route mail to me? He probably doesn't know exactly where
NJ is, and certainly not where Emerson is, but he's got a way to
"pass the buck". He sends his letter on to a "bigger" post office,
and that office sends it to a bigger one until they find someone
that's heard of NJ. (Keep your snide remarks to yourself!) Let's say
that that post office is in London. London would then know that in
order to get that mail to NJ, it has to send it to a particular site
in NJ (let's say it's in Trenton, though I belive that it's actually
acceptable to have several of these types of sites.) That site
would then know directly how to get to Emerson (which might be by
using the other style paths above). The Trenton post office is
something that I believe is called a "domain server", since it knows
about all the subdomains under it. From that point on, delivery goes
as normal.

As with everything, this gets more complicated in real life, because
not everyone is in an officially registered domain. For instance,
although I'm working on getting a "local" mail feed for a small unix
machine I use at work, currently the only way to get to me more-or-
less reliably is through Paul Homchick's system, "cgh". His system
isn't registered either, but he does have a local connection to a
site who is in an officially registered domain. What I can then do
for a mailing address is to give a path through these systems to
mine, using a combination of the domain-type address and the path-
type address, so my address at work is:

        cgh!monymsys![e-mail]

That says (more or less) that I can be reached by:
        1) Go to the US domain, then:
        2) Go to the PA (Pennsylvania) domain, then:
        3) Go to the PHA (Philadelphia) domain, then:
        4) Go to "manta" (which is a single machine), then:
        5) Go to "cgh" (a single machine), then:
        6) Go to "monymsys" (my machine) and deliver to david.

I've left out a lot of the actual mechanics, but this message has
gotten too long already. In any case, I hope this proves useful, and
I'd appreciate any corrections from those who understand this better
than I do.

Addendum: Since I wrote this message, my own connectivity has improved
greatly. To give another example as above, I can now be reach at:

       monymsys![e-mail]

As with the above address, I'll break it down:
        1) Go to the .COM domain server, then:
        2) Go to the registered domain known as "vmp", then:
        3) Have vmp forward that main to monymsys and deliver to david.

There are a few interesting things here that make this a better example
than the initial one. In my first example, there's a strong notion
of physical location. This is actually a misleading example, since
in reality, there are relatively few location-based domains and many more
"organizational" domains. Note that there's no explanation of who or where
the domain server for the .COM domain is; that's exactly the whole point.
If you send mail from a machine that's not the domain server for the
.COM domain (and there aren't that many of them), your machine will know
either how to get to such a server, or at least how to get to a "smarter"
computer (which might pass your mail on to a yet smarter machine, and so
on).

There's an awful lot more to it than what I've explained. I wrote this
message initially around a year ago, and although I've learned a lot
more about how domains work (and hopefully I'll have my machines at
work registered within the .COM domain soon), the thing that I've learned
the most is that there's a LOT of information to know out there!

I hope you've found this discussion helpful. If I can help, please feel
free to drop me a line at one of the addresses listed at the beginning
of this file.