On Tue, 9 Jun 1998, Jay R. Ashworth wrote:
Hierarchicality is almost forced by the architectural design of the current implementation of DNS; and I got a hot scoop for you: you won't get a flag day on DNS.
The only hierarchy that is forced by DNS is the hierarchy of delegation of subdomains and the hierarchy of search paths to resolve a name to an address. The DNS does not favor any particular categorization scheme. You can just as easily use www.ca.example.com and www.us.example.com as you can use www.example.us and www.example.ca.
Dream on. DNS is an addressing scheme just like "123 Any St., Anytown, USA". It does a job that needed to be done, more or less well. If you want something different then find people who will pay for it and build it. I suspect you will find that there is little demand and no money available to build a universal index of everything there is.
It would be you, would it not, who "wants something different"?
No. The naming schemes that people currently apply to the DNS are diverse and chaotic. I don't want to see that changed by imposing a top-down set of rules on the DNS and that is what everyone else in this thread has suggested. It matters not if there was some historical understanding that was followed by the DNS registry 6 years ago. That's not how things work now. And so far as I can see the Magaziner white paper has given the green light to IANA to go ahead and expand the top level namespace in an orderly fashion. It is unlikely that they will impose any rules on the end-users of domain names, only on the registry system itself to ensure that things proceed in an orderly fashion.
You're correct, making DNS into anything except a very coarse index is infeasible. But I don't see any reason to specifically _avoid_ using DNS as at least a classification tool so people know what to expect when they go somewhere.
Here's one reason. Because it is impossible to make DNS into anything but a very coarse index. I remember seeing a documentary of a scientist in Florida that was studying alligators. He was using a computer to record and analyze his data. The video showed him entering data into a DOS machine running EDLIN and then using GWBASIC to process the data. Other people will swear that you need ORACLE and Mathematica to do scientific data analysis and recording. If you have a nail that needs hammering then every tool looks like a hammer. I took my kids to a birdhouse building class this spring and there weren't enough hammers to go around so I went out into the parking lot and got a rock. It did the job and they were pleased to learn that you can hammer nails without a hammer. But if I were recommending tools for a birdhouse-building factory, you can be sure I would not recommend hammers. Thus the fact that DNS *COULD* be used as an indexing system is irrelevant. The real question is: given that the Internet would greatly facilitate the use of an indexing system, how could one best be built? And I think there is a real answer to this question that could be discovered if enough folks would pull together an IETF working group that includes some librarians and some protocol designers.
We're veering far off-topic for NANOG here, quick; let's get back on topic before everyone flies home. :-)
Nobody but you and I are reading this thread anymore. -- Michael Dillon - Internet & ISP Consulting Memra Communications Inc. - E-mail: michael@memra.com http://www.memra.com - *check out the new name & new website*