North American Network Operators Group|
Date Prev | Date Next |
Date Index |
Thread Index |
Author Index |
Re: [NANOG] Re: new.net: yet another dns namespace overlay play
- From: David Charlap
- Date: Thu Mar 08 13:02:24 2001
Vadim Antonov wrote:
> Build a search engine which takes "old" domain name "WWW.CNN.COM" and
> produces URL with 188.8.131.52 in it :)
And then build wedges into operating systems and application software to
automatically access this search engine when presented with names.
Congratulations, you've just re-invented DNS, but with more overhead.
No matter what you end up using, you are going to need some kind of
directory service for users to look up the entities they are accessing.
If DNS doesn't cut it for some applications, then perhaps another system
can be added. One such example is the "Internet keywords" concept used
by AOL and some web browsers. (See also http://www.realnames.com/)
This doesn't require that the existing DNS system be trashed.
> I'm wondering how people managed to find CNN on TV -- after all, CNN
> ads didn't feature local channel numbers :)
Most cable companies send customers a printed list of channel mappings,
and mail out updates when the lineup changes. Or they just channel-surf
and look at every single channel until they find it.
Telephone systems (which are closer in magnitude to the internet than
television systems) send customers printed directories every year and
provide (usually for a fee) a directory assistance service.
None of these systems work any better than DNS:
- A quick reference card can't work when you've got millions of hosts.
- Channel surfing is equally useless. The internet is several orders
of magnitude bigger than a television system.
- Printed directories are impractical and expensive. Given the rate
of change, you'd need to reissue it at least once a month. Putting
the directory on CD-ROM may help with the publication costs, but you
would still need to distribute them. And customers will have to pay
for a subscription to this. And you'd still need a directory service
in order to allow access to nodes that are added/changed and have
not yet gotten into the directory.
If you end up relying on an on-line service, then you've just