We are planning to delegate every county name as a locality name in each State. For example, "Ventura.CA.US", "Dade.FL.US", "Riverside.CA.US".
Too many levels is _very_ bad, as people will stick to using old three-letter domains. How about limiting the scheme to <orgname>.<state>.us? Or, even better, for major metropolitan areas like SF, LA, NYC, CHI etc the name central city can be used instead of state. That scheme worked reasonably well for .SU :) --vadim
On Wed, 26 Jul 1995, Vadim Antonov wrote:
Or, even better, for major metropolitan areas like SF, LA, NYC, CHI etc the name central city can be used instead of state.
That scheme worked reasonably well for .SU :)
But everyone knows that in comparison to .US, .SU was a backwards place ;-) Michael Dillon Voice: +1-604-546-8022 Memra Software Inc. Fax: +1-604-542-4130 http://www.memra.com E-mail: michael@memra.com
Or, even better, for major metropolitan areas like SF, LA, NYC, CHI etc the name central city can be used instead of state.
I like this idea. (Note that VIXIE.SF.CA.US has moved many times since its creation, and though I've always been in the Bay Area I've not lived in San Francisco since 1988.)
On Wed, 26 Jul 1995, Paul A Vixie wrote:
Or, even better, for major metropolitan areas like SF, LA, NYC, CHI etc the name central city can be used instead of state.
I like this idea. (Note that VIXIE.SF.CA.US has moved many times since its creation, and though I've always been in the Bay Area I've not lived in San Francisco since 1988.)
This would seem to indicate that geographical domains are a bad idea and that domain names should be based on some characteristic that is less likely to change over time. VIXIE.BIND.HACKER ? Michael Dillon Voice: +1-604-546-8022 Memra Software Inc. Fax: +1-604-542-4130 http://www.memra.com E-mail: michael@memra.com
I like this idea. (Note that VIXIE.SF.CA.US has moved many times since its creation, and though I've always been in the Bay Area I've not lived in San Francisco since 1988.)
This would seem to indicate that geographical domains are a bad idea and that domain names should be based on some characteristic that is less likely to change over time.
VIXIE.BIND.HACKER ?
Given the tremendous demand for short, "sexy" domain names, and that we only have 70,000 companies in .COM out of 25,000,000 mid-to-large-sized businesses in the United States, we are indeed fast approaching the point where domain names will no longer map meaningfully to the objects they identify. Something like .US which is currently for individuals will have an even tougher time growing to 200,000,000+ individuals. I've been kicking around VIXIE.FAM et al, on the assumption that the first use of a name under .FAM (family) would be responsible for setting up the tree for the rest of the folks using that name. On the other hand, SMITH.FAM would be a pretty huge undertaking and I'm leaning more toward something that service providers could do as a third-party. I don't know what that is yet. I wish that Padlipski had not retired - we collectively need his wisdom. "The map is not the territory." No hierarchy will map easily to all registrants -- the goal is to find something that will work, no matter how painful it is, and let the directory services people handle the mapping of real-world object names like countries and cities and families and companies, into funny-world objects like host names and URLs and so forth. Assigning hexadecimal strings at random would be better than what we have now -- in terms of scalability to the next order of magnitude in network size -- just so long as the strings were unique. Sort of the "social security number" concept only on a wider scale. That said, ".HACKER" would probably not be a useful top level domain given that the majority of DNS registrants will not be computer or network professionals for very much longer. So far I'm headed toward "Label.Hash.COM.US" where Label is something like SUN or IBM or VIX, Hash is a variable sized token generated from Label and intended to keep the single .COM.US domain from growing into a monster. "Label.Hash.COM.State.US" is also a possibility, that's up to the USDOMREG. Closing .COM and moving to this new structure is going to be a huge undertaking, of course.
On Wed, 26 Jul 1995, Paul A Vixie wrote:
This would seem to indicate that geographical domains are a bad idea and that domain names should be based on some characteristic that is less likely to change over time.
VIXIE.BIND.HACKER ?
On the other hand, SMITH.FAM would be a pretty huge undertaking and I'm leaning more toward something that service providers could do as a third-party. I don't know what that is yet.
I think the key is to let people choose. The county name system and the .FAM system do not give people a choice. If cousin Ned wants to be in the .KLINGON domain and Aunt Sarah wants to be in the .SEWMISTRESS domain, why not let them? Realistically, in order to keep the root under control these should be .KLINGON.ROLE and .SEWMISTRESS.ROLE but maybe Aunt Martha is happy with the .SMITH.FAM domain. I think the better solution will com out of a judicious expansion of the root domains to allow for international choices similar to what .COM, .NET and .ORG already provide.
let the directory services people handle the mapping of real-world object names like countries and cities and families and companies, into funny-world objects like host names and URLs and so forth. Assigning hexadecimal strings at random would be better than what we have now -- in terms of scalability to the next order of magnitude in network size -- just so long as the strings were unique. Sort of the "social security number" concept only on a wider scale.
Now I am one of those unusual people who knows their number off by heart which is good since the card has long since disappeared into a Northern river back in 1979. However, I would not wish to break the substantially mnemonic tradition of DNS. If we are going to have funny-world objects, lets allow people to be creative rather than forcing a mapping scheme on them. Lets not forget that domain names are "names". As such, they should be as rich and vibrant as real-world names. In the real world we have Mister T, Prince, and Wim Vandeneerenbeemt, why can't the funny-world have these kind of names too.
That said, ".HACKER" would probably not be a useful top level domain given that the majority of DNS registrants will not be computer or network professionals for very much longer.
Right. Top levels should be reserved for general categories. It should be VIXIE.BIND.HACKER.ROLE
So far I'm headed toward "Label.Hash.COM.US" where Label is something like SUN or IBM or VIX, Hash is a variable sized token generated from Label
Those 5 digit FIPS county codes sound like a good hash. Even if they didn't exist, just get a list of US counties, number them and there you are, hash!
Closing .COM and moving to this new structure is going to be a huge undertaking, of course.
.com is international. It will never be closed. However an attractive .com subgrouping system within .us could entice many businesses with regional or local focus to go for .com.us or .com.state.us or .com.county.us Michael Dillon Voice: +1-604-546-8022 Memra Software Inc. Fax: +1-604-542-4130 http://www.memra.com E-mail: michael@memra.com
Michael Dillon writes:-
I think the key is to let people choose. The county name system and the .FAM system do not give people a choice. If cousin Ned wants to be in the .KLINGON domain and Aunt Sarah wants to be in the .SEWMISTRESS domain, why not let them? Realistically, in order to keep the root under control these should be .KLINGON.ROLE and .SEWMISTRESS.ROLE but maybe Aunt Martha is happy with the .SMITH.FAM domain.
Hmm. If you are looking for ideas (who said that they had to be good? what's good about a geographic/city split anyway?), why not have <SURNAME>.US, and let the family tree start to hang together. If there are, say, too many DILLON.US branches of the family, then the Dillons themselves can split this by whatever method suits them. If there are strong ties with those Dillons who arrived in 1687, then 1687.DILLON.US could be used, if the Dillons in Boston form a unit, then BOSTON.DILLON.US and so what if someone moves out of Boston, they still have the same family relationship. Mike -- Mike Lawrie <mlawrie@apies.frd.ac.za> Manager: Uninet Ph: +27 12 841 3542 Foundation for Research Development Fx: +27 12 804 2679 P O Box 2600, Pretoria 0001, RSA ^^ +27 12 349-1179 sometime in 1995
what's good about a geographic/city split anyway?), why not have
Top level geographies are most likely to map to organizations willing to maintain subdelegations. Other than that, it's all arbitrary other than that it's necessary to have a deep tree. I already know that U.S. companies put under states are going to feel overspecified. I can already hear the reasons why IBM.COM.NY.US is supposedly "wrong" since IBM is a nationwide, no, worldwide company. On the other hand they're incorporated in New York and the point isn't to have the name make sense, it's to keep the hierarchy from flattening. When 25,000,000 companies have domain names, there will be at least three labels in most of those top level names. It cannot be helped.
Date: Thu, 27 Jul 1995 07:42:05 -0700 From: Paul A Vixie <paul@vix.com>
I already know that U.S. companies put under states are going to feel overspecified. I can already hear the reasons why IBM.COM.NY.US is supposedly "wrong" since IBM is a nationwide, no, worldwide company. On the other hand they're incorporated in New York and the point isn't to have the name make sense, it's to keep the hierarchy from flattening.
To make matters far worse, a very disproportionate number of corporations are "located" in Delaware for legal and tax reasons, even if they only have a small office in that state. Whoever is responsible for DE will have a very big job! And think of all the traffic that will cross teh pond because people forget the ".US"! I think Paul is right on the mark on this one. R. Kevin Oberman Energy Sciences Network (ESnet) National Energy Research Supercomputer Center (NERSC) Lawrence Livermore National Laboratory (LLNL) EMAIL: koberman@llnl.gov Phone: +1 510 422-6955
On Thu, 27 Jul 1995, Paul A Vixie wrote:
what's good about a geographic/city split anyway?), why not have
Top level geographies are most likely to map to organizations willing to maintain subdelegations. Other than that, it's all arbitrary other than that it's necessary to have a deep tree.
I already know that U.S. companies put under states are going to feel overspecified. I can already hear the reasons why IBM.COM.NY.US is supposedly "wrong" since IBM is a nationwide, no, worldwide company.
In the .ca domain, companies with offices in two or more provinces can have a top level name, i.e. ibm.ca. With offices in two or more cities in a single province they can have a name at the next level, i.e. widget.ab.ca. If they are in one city only, their name must include the city as in joes-eats.vancouver.bc.ca.
When 25,000,000 companies have domain names, there will be at least three labels in most of those top level names. It cannot be helped.
You could use a system similar to .ca which allows you to remove nonsense geographic labels but still elegantly handles the millions of small businesses with presence in a single town. Michael Dillon Voice: +1-604-546-8022 Memra Software Inc. Fax: +1-604-542-4130 http://www.memra.com E-mail: michael@memra.com
One thing to keep in mind is that not all of .com is broken; major country and world wide companies should sensibly have a short address, we just need an adaptive mechanism to allow more local companies to come along without losing all usability within the existing .com structure... I wouldn't have bothered to bring this up, but someone mentioned using an intermediate hash on *all* .com's (ibm.49.com), which is a waste of time. Every extra level of address hurts a bit, and it makes sense that "really big" companies should have simple .com addresses. We don't have to retrofit the new mechanism to everyone for consistency's sake. george william herbert gherbert@crl.com KD6WUQ Unix / Internet Consultant http://www.crl.com/~gherbert
[...] I wouldn't have bothered to bring this up, but someone mentioned using an intermediate hash on *all* .com's (ibm.49.com), which is a waste of time. Every extra level of address hurts a bit, and it makes sense that "really big" companies should have simple .com addresses.
That's sensible engineering practice and in other circumstances I would applaud it. However, we are not in a sensible engineering situation or any other kind of sensible situation -- people, and especially lawyers and marketeers are involved. People currently worry about and choose domain names with the same kind of intensity they experience when worrying about and choosing trademarks, product names, company names, and logos. The domain name is the "service mark" of the 90's -- people "do business under it" and its easy recognition by customers is considered valuable by the folks who steer multinational companies. Currently it is possible for a little company to look "just like" a big company. They all appear with the same .COM suffix, in the same whois registries, and just as noone knows you are a dog, noone knows when they see e-mail from you that you're just a one-person consulting shop or whatever. Witness the MTV.COM debacle, or find out which of the hundreds of daily newspapers with "Examiner" in their name is registered as "examiner.com". We have _got_ to anhililate the value of these names. There is no way on god's green earth that a small company is going to allow themselves to be put way down in a backwater while the more visible namespaces are available to bigger companies. These people will lie on their applications, they will find out what the categorization criteria is and pretend to be something they are not, they will cause the NIC and any other registries to spend a great deal of time trying to verify this information, and ultimately when all is said and done and they don't get what they want, they will _sue_ for restraint of trade. Everyone, large and small, has to be treated as equally as possible. And the domain names have to be quite a lot uglier than they are now, such that the tendancy to register under .NET,.COM,.ORG,etc just to protect the company name or trademark(s) will no longer bear useful fruit for those who do it. That said, I am not in favour of making these names so long that they are not usable. Larger organizations tend to have deeper interior DNS trees, and it would be Really Bad if we end up with four labels just to get from the root to the organization, and another four levels to get from the root of the organization to some host within the organization. Perhaps this indicates a need to make RFC1535 a _requirement_ for Internet hosts, so that organizations can use search lists and partially qualified interior names, thus isolating them from the organization's depth in the external namespaces. This feels like a slippery slope and I'm not sure I want to pursue it just yet. Two-label names don't scale. Three-label names can be made to do so -- and I would give in on the ".Hash." component of my previous proposal if I had a good idea for a second-label that would cause full and healthy looking trees. ".State.US" has the advantage that the USDOMREG could ask the various state governments to take responsibility for third-level registration, much as they do for corporation names now. I agree that once you're down into .City.State.US or .County.State.US, it is no longer feasible for an organization with even a moderately deep interior tree to register. Six-label fully qualified names aren't usable since they are no longer a syntactic improvement over raw addresses. (They are a slight semantic improvement, since they'll change less often. I don't know yet whether that matters enough.)
That's sensible engineering practice and in other circumstances I would applaud it. However, we are not in a sensible engineering situation or any other kind of sensible situation -- people, and especially lawyers and marketeers are involved.
Oh lordy, not the lawyers...
[...] We have _got_ to anhililate the value of these names. There is no way on god's green earth that a small company is going to allow themselves to be put way down in a backwater while the more visible namespaces are available to bigger companies. These people will lie on their applications, they will find out what the categorization criteria is and pretend to be something they are not, they will cause the NIC and any other registries to spend a great deal of time trying to verify this information, and ultimately when all is said and done and they don't get what they want, they will _sue_ for restraint of trade.
There's a wonderful, much more simple solution... charge for .com addresses on a fee scale which will discourage companies with under $X/yr business from getting straight .com addresses. As a possible example, $25k/yr for a .com domain name is going to keep most 5-20 person companies out of it (unless they REALLY want it bad) but won't be noticable to most 100 person companies and Sun and IBM. And $25k/site would fund a lot of NIC activity 8-) If you want to be aggressively egalitarian about it, apply it retroactively (which will put my vanity domain out of action, but what the hey). A solution along these lines will let us keep a better engineering-side solution without having the customer or NIC have to jump through hoops to determine validity of applications. I think that rabidly renaming everyone would be overkill. The problem is not (basically) the existing .com, but the future million-plus-businesses case. Establishing a policy now which manages the growth of the .com domain by discouraging its use, but not eliminating it, is a good thing. But we don't have to get all our customers angry at us all at once to do that... we just have to convince most of the new customers that they really don't want a .com address unless they *really* want it... -george george william herbert gherbert@crl.com KD6WUQ Unix / Internet Consultant http://www.crl.com/~gherbert
There's a wonderful, much more simple solution... charge for .com addresses on a fee scale which will discourage companies with under $X/yr business from getting straight .com addresses. As a possible example, $25k/yr for a .com domain name is going to keep most 5-20 person companies out of it (unless they REALLY want it bad) but won't be noticable to most 100 person companies and Sun and IBM. And $25k/site would fund a lot of NIC activity 8-)
A charging model for this makes a certain amount of sense, but there are still some problems. First, the IANA is not really empowered to collect fees, and the NIC (who is funded to maintain R&E domains, not commercial ones) takes NSF money and would therefore be in a difficult position if they charged fees not clearly delineated in its contract. In other words this is an ISOC/IESG/IAB problem, and a hard one at that. This is the kind of thing that's best done by (dare I say it?) a 501c6 nonprofit trade association. (Ahem.)
If you want to be aggressively egalitarian about it, apply it retroactively (which will put my vanity domain out of action, but what the hey).
batmanforever.com is a vanity domain. I suspect that yours is not one.
A solution along these lines will let us keep a better engineering-side solution without having the customer or NIC have to jump through hoops to determine validity of applications.
That's certainly true.
I think that rabidly renaming everyone would be overkill. The problem is not (basically) the existing .com, but the future million-plus-businesses case. Establishing a policy now which manages the growth of the .com domain by discouraging its use, but not eliminating it, is a good thing. But we don't have to get all our customers angry at us all at once to do that... we just have to convince most of the new customers that they really don't want a .com address unless they *really* want it...
If we had used that argument the last time around, there would still be hundreds of .ARPA "domains" floating around. I'm against it, partly because it violates the least astonishment principle (when guessing at a domain, one would have to wonder whether the organization was named under the ancient or recent rules), and partly because .COM as presently populated is terribly ugly. The idea I've been kicking around would be relatively painless and should not make very many people very angry all at the same time (that is, I'm all for spreading the anger out across a longer schedule.) 1. develope an alternative to .COM, get its infrastructure in place. 2. close .COM to new entries, point folks at #1, set date for #3. 3. replace all existing .COM entries with nonterminal CNAME's to entries under #1's scheme. delete anything not renewed under #1. set date for #4. 4. remove .COM. I expect the delay between #2 and #3 to be about two years. I expect the delay between #3 and #4 to be about two years. We might be able to get it done before the turn of the century.
There's a wonderful, much more simple solution... charge for .com addresses on a fee scale which will discourage companies with under $X/yr business from getting straight .com addresses. As a possible example, $25k/yr for a .com domain name is going to keep most 5-20 person companies out of it (unless they REALLY want it bad) but won't be noticable to most 100 person companies and Sun and IBM. And $25k/site would fund a lot of NIC activity 8-)
A charging model for this makes a certain amount of sense, but there are still some problems. First, the IANA is not really empowered to collect fees, and the NIC (who is funded to maintain R&E domains, not commercial ones) takes NSF money and would therefore be in a difficult position if they charged fees not clearly delineated in its contract.
In other words this is an ISOC/IESG/IAB problem, and a hard one at that. This is the kind of thing that's best done by (dare I say it?) a 501c6 nonprofit trade association. (Ahem.)
I see US centric ideas creaping in again. .COM is not a US specific problem and US specific solutions will not fly. The fiscal argument may involve a common demomination... say the EUMU. As to the closing of .COM... why stop there? Why not shutdown all the non geographic aligned domains? People will just flee to .INT, .ORG and other locations. Either that or establish arbitrary rules, along the lines that were established for .EDU, .MIL, .GOV and all the other country TLDs. I see no reason to close .COM. --bill
There will be a charge for each .com domain name BEFORE the end of this year... and not thank the lord $25,000 per domain name. I'll say more when saying more becomes appropriate. ******************************************************************** Gordon Cook, Editor & Publisher Subscript.: Individ-ascii $85 The COOK Report on Internet -> NREN Non Profit. $150 431 Greenway Ave, Ewing, NJ 08618 Small Corp & Gov't $200 (609) 882-2572 Corporate $350 Internet: cook@cookreport.com Corporate. Site Lic $650 http://www.netaxs.com/~cook <- Subscription Info & COOK Report Index ********************************************************************
Paul A Vixie previously wrote:
That's sensible engineering practice and in other circumstances I would applaud it. However, we are not in a sensible engineering situation or any other kind of sensible situation -- people, and especially lawyers and marketeers are involved.
People currently worry about and choose domain names with the same kind of intensity they experience when worrying about and choosing trademarks, product names, company names, and logos. The domain name is the "service mark" of the 90's -- people "do business under it" and its easy recognition by customers is considered valuable by the folks who steer multinational companies.
Currently it is possible for a little company to look "just like" a big company. They all appear with the same .COM suffix, in the same whois registries, and just as noone knows you are a dog, noone knows when they see e-mail from you that you're just a one-person consulting shop or whatever. Witness the MTV.COM debacle, or find out which of the hundreds of daily newspapers with "Examiner" in their name is registered as "examiner.com".
One doesn't need a two-level domain name to help make a small business look bigger.
We have _got_ to anhililate the value of these names. There is no way on god's green earth that a small company is going to allow themselves to be put way down in a backwater while the more visible namespaces are available to bigger companies. These people will lie on their applications, they will find out what the categorization criteria is and pretend to be something they are not, they will cause the NIC and any other registries to spend a great deal of time trying to verify this information, and ultimately when all is said and done and they don't get what they want, they will _sue_ for restraint of trade.
Everyone, large and small, has to be treated as equally as possible. And the domain names have to be quite a lot uglier than they are now, such that the tendancy to register under .NET,.COM,.ORG,etc just to protect the company name or trademark(s) will no longer bear useful fruit for those who do it.
What sort of horseshit is this? Listen, domain names _have_ value. You can't avoid it. But there's the problem of crowded name spaces, which is not an impossible challenge technically speaking, but which limits the choices of names that those who don't have one yet can get. The solution? Create various domains at various levels, make contracts for the right to register a name under one of these domains, and sell the contracts in the open market. The larger companies are the ones that will be able to _afford_ the two-level names; there's no need to verify money! [Well, ok, you have to make sure the money ain't fake, but that's easy :)] Think about it: if IBM wants IBM.COM, rather than IBM.COM.NY.US, then let them buy the right to register a name in the .COM hierarchy, but they should pay more for that than for a name in the COM.NY.US. Seriously, in the long run, many services will be tradeable as comodities in futures markets; I know people who are trying to do this now with bandwidth, and hell, there are people making small fortunes by charging to register and "hold" domain names for others. Yet we still have a lousy excuse for a NIC that _DOES NOT_ charge for its services!!! Get with it. Capitalism, it'll fix the problem at hand. Anyways, yes, what I propose does have one sticky point: who should charge for the registration services and what is to be done with that money; any ideas? [...]
I would give in on the ".Hash." component of my previous proposal if I had a good idea for a second-label that would cause full and healthy looking trees. ".State.US" has the advantage that the USDOMREG could ask the various state governments to take responsibility for third-level registration, much as they do for corporation names now.
The damn governments haven't even gotten into any of this folks! And there's nothing wrong with two-level names so long as people can still come up with unique two-level names; it is quite feasible to have domain name servers for hundreds of thousands of names in one hierarchy, those who say it isn't are either lazy or incompetent or don't know what they are talking about (and I'm talking about good, reliable, fast servers here). It is the fact that it gets hard to find the name you want not taken when you have hundreds of thousands of names in one hierarchy that will force us away from the current system. And BTW, me and my business partner were looking for a domain name for a soon to be business we want to start, and quickly found that our first four choices of a name were already taken. That is the future unless we do something about it.
I agree that once you're down into .City.State.US or .County.State.US, it is no longer feasible for an organization with even a moderately deep interior tree to register. Six-label fully qualified names aren't usable since they are no longer a syntactic improvement over raw addresses. (They are a slight semantic improvement, since they'll change less often. I don't know yet whether that matters enough.)
Anyways, there aren't that many businesses, and not a significant fraction of the 260e6 + people in the US are likely to try to register domains of their own: the vast majority will be happy enough with a mailbox on someone else's domain (like, their ISP's, as they do now; we may have 30e6 users on the net, but the number of domains registered to American individuals + American companies is orders of magnitude smaller). There's no need to introduce some sort of communism of the net and talk about "anihilating" the value of names. Please, the thought turns my stomach. Nick
On Fri, 28 Jul 1995, Nicolas Williams wrote:
Think about it: if IBM wants IBM.COM, rather than IBM.COM.NY.US, then let them buy the right to register a name in the .COM hierarchy, but they should pay more for that than for a name in the COM.NY.US. Seriously, in the long run, many services will be tradeable as comodities in futures markets;
In Canad we have an agricultural commodity known as "quota" which is sold on the open market. There are provincial marketing boards, like the Ontario Milk Marketing Board, which have set a quota on the number of litres of raw milk that may be produced by farmers in the province. This annual quota is divided up and allocated to farmers. Originally it was a cheap thing to get, but you could only get so much quota. If your herd produced 400 gallons a day and the board would only issue you with 300 gallons of quota, you could dump the milk, sell the cows, feed the milk to pigs, but you could not sell that milk, period. So some farmers figured out that it was worth a considerable sum to buy quota from other farmers and that they could make money by increasing their dairy herds to the carrying capacity of their farms. Nowadays it is not unusual for quota to be sold for the value of 2 years production which means the farmer doesn't start to make money on the quota for two years. Nobody intended for this to happen, but it did happen. If the INternic does not loosen up the bonds on top level domains it will happen in this industry too. I think top level domains should be issued like the FCC issues frequencies. Some should be free like .COM, .NET, .ORG and some should be auctioned off like .PSI, .MCS, .NETCOM, .UUNET etc. Hold an auction for three top level domains to the highest bidder. Winners get to choose their name greater than 2 characters (ISO 3166) and less than 8 characters. Next year auction off another three. At each auction, issue a few free toplevels as well. How many companies would pay to have yourname.msn for their domain name?
what I propose does have one sticky point: who should charge for the registration services and what is to be done with that money; any ideas?
ISOC should do this. Hand over the Internic to ISOC, get them to charge a small fee for transactions that will cover all expenses plus allow for improvements in registry and directory services such as using modern database technology to manage the system.
come up with unique two-level names; it is quite feasible to have domain name servers for hundreds of thousands of names in one hierarchy, those who say it isn't are either lazy or incompetent or don't know what they are talking about (and I'm talking about good, reliable, fast servers here).
Modern database technology on striped disk arrays can handle the entire world's population using a 1 level domain; pick a number between 1 and 6 billion. Michael Dillon Voice: +1-604-546-8022 Memra Software Inc. Fax: +1-604-542-4130 http://www.memra.com E-mail: michael@memra.com
On Thu, 27 Jul 1995, Paul A Vixie wrote:
We have _got_ to anhililate the value of these names.
When industrial real-estate is hard to find, prices go WAY UP! But when the city zoning committee rezones land to open up a new industrial park, the value of the average industrial real-estate drops. What do you think will happen when the Internic allows .BIZ, .INC, .LTD, .PTY, .GMBH, .OY, and .CORP to be used? Of course, IBM will be able to prevent anyone else from registering as IBM.CORP simply because they own the trademark. But widening the roor namespace combined with a rule that only INCORPORATED companies can register in the international commercial domains will solve many of these problems. green.com, green.corp, green.inc, green.biz and green.ltd will be completely different companies but they all get nice short names. And the folks at Green Auto Parts in Raleigh NC can still register as green.com.nc.us and they will be quite happy since they only have 10 machines on their store LAN.
god's green earth that a small company is going to allow themselves to be
Yep, we need a .GOD and a .EARTH domain as well. Who will register TRUE.GOD with a machine named ONE as their web server?
Everyone, large and small, has to be treated as equally as possible. And the domain names have to be quite a lot uglier than they are now, such that the tendancy to register under .NET,.COM,.ORG,etc just to protect the company name or trademark(s) will no longer bear useful fruit for those who do it.
This is always a problem when a resource is made artificially scarce. Michael Dillon Voice: +1-604-546-8022 Memra Software Inc. Fax: +1-604-542-4130 http://www.memra.com E-mail: michael@memra.com
but they all get nice short names. And the folks at Green Auto Parts in Raleigh NC can still register as green.com.nc.us and they will be quite happy since they only have 10 machines on their store LAN.
but they won't. they'll want green.biz. if you were right that they'd be quite happy, then our local newspaper would have called their domain SF-EXAMINER.COM rather than EXAMINER.COM. but they were almost certainly worried that if they didn't take EXAMINER.COM, some other daily newspaper would surely take it, and they'd somehow be worse off. we're getting well into the murky arena where perception begins to dominate the equation. people _believe_ that short domain names indicate quality or size or "greatness" (i call this "domain envy"), and so it becomes true. it's sort of like thin-vs-wide ties, or hemlines. i'm not sure that we as technologists are well equipped to deal with this issue, but i am quite certain that the nontechnologists would do an even worst job, so here we are.
On Thu, 27 Jul 1995 19:56:35 -0700 Paul A Vixie wrote:
Everyone, large and small, has to be treated as equally as possible. And the domain names have to be quite a lot uglier than they are now, such that the tendancy to register under .NET,.COM,.ORG,etc just to protect the company name or trademark(s) will no longer bear useful fruit for those who do it.
As far as I know, there is no 'congestion' on the 800-namespace. maybe, one of the reasons is that these 'records' have a fixed, 7-digit length, and 'vanity' numbers have the same length as ugly ones, spreading the load. Maybe we should enforce a minimum length on any complete domain name? Geert Jan
As far as I know, there is no 'congestion' on the 800-namespace. maybe, one of the reasons is that these 'records' have a fixed, 7-digit length, and 'vanity' numbers have the same length as ugly ones, spreading the load.
Currently, the 800 space is filling up and they're considering using another 3 digit area-code (hadn't heard on what they'd finally decided on) for toll-free calls. The curious thing is that if you have a vanity number in 800, there's some motivation to get the same vanity number under the new area code. Doesn't help the {un}clogging of 800 at all... maybe the IANA would be willing to sell vanity IPv6 addresses :-). henry
On Fri, 28 Jul 1995, Geert Jan de Groot wrote:
As far as I know, there is no 'congestion' on the 800-namespace. maybe, one of the reasons is that these 'records' have a fixed, 7-digit length, and 'vanity' numbers have the same length as ugly ones, spreading the load.
actually, I think there is. that is why they are opening up the '888' area code to be toll free also.
Date: Fri, 28 Jul 1995 08:26:34 -0700 (PDT) From: Nikos Mouat <nikm@ixa.com>
On Fri, 28 Jul 1995, Geert Jan de Groot wrote:
As far as I know, there is no 'congestion' on the 800-namespace. maybe, one of the reasons is that these 'records' have a fixed, 7-digit length, and 'vanity' numbers have the same length as ugly ones, spreading the load.
actually, I think there is. that is why they are opening up the '888' area code to be toll free also.
And they are seeing the same problems. 1. No one wants to be in 888. There is a belief consumers will not understand that 888 is a functional equivalent to 800 for a long time and 888 numbers will be at a disadvantage. 2. Holders of vanity numbers in 800 space are demanding that they get the same numbers in 888 space! They say that the vanity numbers are functional trademarks and that they (whoever "they" are) will sue if they don't get the 888 number as well as the 800. If this does not sound familiar, you have not been reading your mail lately! R. Kevin Oberman Energy Sciences Network (ESnet) National Energy Research Supercomputer Center (NERSC) Lawrence Livermore National Laboratory (LLNL) EMAIL: koberman@llnl.gov Phone: +1 510 422-6955
In message <199507272214.AA07015@mail.crl.com>, George Herbert writes:
One thing to keep in mind is that not all of .com is broken; major country and world wide companies should sensibly have a short address, we just need an adaptive mechanism to allow more local companies to come along without losing all usability within the existing .com structure...
I wouldn't have bothered to bring this up, but someone mentioned using an intermediate hash on *all* .com's (ibm.49.com), which is a waste of time. Every extra level of address hurts a bit, and it makes sense that "really big" companies should have simple .com addresses.
We don't have to retrofit the new mechanism to everyone for consistency's sake.
george william herbert gherbert@crl.com KD6WUQ Unix / Internet Consultant http://www.crl.com/~gherbert
Is ibm.nyse.com, dec.nyse, mci.nasdaq.com a problem? This takes the arbitrary decision out of accepting or denying someone access to the .com hierarchy and enforces and already strict naming. US citizens would have to remember country codes for Germany to get Seimens, Holland to get Phillips, etc. Not a terrible side effect. Some US citizens may even come to startling revelations like that Denmark, Hollans, Norway, and Sweeden are separate countries. :-) For US companies everyone would need to know the exchange and trading symbol for US companies or what state the company was in if privately held. For most big US corporations nyse.com is a good bet, although some of the trading symbols are less cute than what we see in .com. There are too many proposals already, but having a non-arbitrary means of denying access to a small set of top levels (nyse.com, amex.com, nasdac.com) is a handy feature. Curtis
On Thu, 27 Jul 1995, George Herbert wrote:
One thing to keep in mind is that not all of .com is broken; major country and world wide companies should sensibly have a short address,
I do not think that is so sensible. One of the things touted about the Internet is that it is a great equalizer for businesses. Small businesses can compete on a global scale, and can have an internet presence identical to mega-companies. By causing a break based solely on "how big are you," I fear a small revolt. We _REALLY_ try to sell to our customers the idea of using widget.ral.nc.us, but they _know_ the .com exists, and if we won't give it to them, they will go elsewhere. Names are everything to businesses. So we have to give in at times, just to stay in business. If a change is made, I feel it needs to be applied globally to all, not just the small businesses. +mike On Thu, 27 Jul 1995, George Herbert wrote:
One thing to keep in mind is that not all of .com is broken; major country and world wide companies should sensibly have a short address, we just need an adaptive mechanism to allow more local companies to come along without losing all usability within the existing .com structure...
I wouldn't have bothered to bring this up, but someone mentioned using an intermediate hash on *all* .com's (ibm.49.com), which is a waste of time. Every extra level of address hurts a bit, and it makes sense that "really big" companies should have simple .com addresses.
We don't have to retrofit the new mechanism to everyone for consistency's sake.
george william herbert gherbert@crl.com KD6WUQ Unix / Internet Consultant http://www.crl.com/~gherbert
Interpath Hostmaster [] hostmaster@interpath.net [] 919-890-6305
If a change is made, I feel it needs to be applied globally to all, not just the small businesses.
that's my feeling as well. on the other hand there are big companies out there who believe that their trademarks and service marks are bound tightly to their domain names. for them we should offer a chance to stay in .COM but for a _very_ high price -- the $25K/year figure quoted here recently wasn't unreasonable. to IBM.COM or DIGITAL.COM this is peanuts, and it will keep their lawyers happy, and it will keep IANA or Internic from being sued for restraint of trade.
On Thu, 27 Jul 1995, Mike Lawrie wrote:
Michael Dillon writes:-
I think the key is to let people choose. The county name system and the .FAM system do not give people a choice. If cousin Ned wants to be in the .KLINGON domain and Aunt Sarah wants to be in the .SEWMISTRESS domain, why not let them? Realistically, in order to keep the root under control these should be .KLINGON.ROLE and .SEWMISTRESS.ROLE but maybe Aunt Martha is happy with the .SMITH.FAM domain.
Hmm. If you are looking for ideas (who said that they had to be good? what's good about a geographic/city split anyway?), why not have <SURNAME>.US, and let the family tree start to hang together. If there are, say, too many DILLON.US branches of the family, then the Dillons themselves can split this by whatever method suits them. If there are strong ties with those Dillons who arrived in 1687, then 1687.DILLON.US could be used, if the Dillons in Boston form a unit, then BOSTON.DILLON.US and so what if someone moves out of Boston, they still have the same family relationship.
I live in Canada, my father came from Ireland. Do we really need DILLON.US, DILLON.CA, DILLON.IE, DILLON.ZA? Wouldn't DILLON.FAM just do as well? Michael Dillon Voice: +1-604-546-8022 Memra Software Inc. Fax: +1-604-542-4130 http://www.memra.com E-mail: michael@memra.com
Given the tremendous demand for short, "sexy" domain names, and that we only have 70,000 companies in .COM out of 25,000,000 mid-to-large-sized businesses in the United States, we are indeed fast approaching the point where domain names will no longer map meaningfully to the objects they identify. Something like .US which is currently for individuals will have an even tougher time growing to 200,000,000+ individuals.
.....
So far I'm headed toward "Label.Hash.COM.US" where Label is something like SUN or IBM or VIX, Hash is a variable sized token generated from Label and intended to keep the single .COM.US domain from growing into a monster. "Label.Hash.COM.State.US" is also a possibility, that's up to the USDOMREG. Closing .COM and moving to this new structure is going to be a huge undertaking, of course.
This idea (that domains can get too large) was the idea behind the bigzone mailing list. John Romkey was the first (to my knowledge) to propose a series of psudo-random labels to create enough heirarchy to support the size and scope of wide scale deployment. Other efforts, to improve the information packing density of the existing labels (wildcards within labels, hierarchy to the right of root) tend to lead to madness. Perhaps it is time to resurect the bigz list for active discussion on how to: identify what a big zone really is how to split a big zone I'm not convinced that .COM needs closing... yet. -- --bill
In message <199507271156.AA14615@zephyr.isi.edu>, Bill Manning writes:
So far I'm headed toward "Label.Hash.COM.US" where Label is something like SUN or IBM or VIX, Hash is a variable sized token generated from Label and intended to keep the single .COM.US domain from growing into a monster. "Label.Hash.COM.State.US" is also a possibility, that's up to the USDOMREG. Closing .COM and moving to this new structure is going to be a huge undertaking, of course.
This idea (that domains can get too large) was the idea behind the bigzone mailing list. John Romkey was the first (to my knowledge) to propose a series of psudo-random labels to create enough heirarchy to support the size and scope of wide scale deployment.
Bill, That a great idea. Just replace the dots in your numeric address with underscore and make it part of the domain name. :-) :-) Then you have to remember that cnidr is in reston and what their numeric address is. I hope people recognize friendly sarcasm. No offense intended Bill. The whole point of a name space is to make things easier to remember. There is no reason you can't be in two domains. Machines can have cnames and people can have multiple MX records. You just have to pick your favorite for the in-addr. Maybe .com can be split by country that you are incorporated in and in the US moved under .us and further split by state of incorporation (or state in which you do primary business for unincorporated businesses and masquerades). Perhaps in the US we could also do symbol.stock-exchg.com.us with a requirement that you be listed on that stock exchange (and maybe even delagated to the exchange), covering publicly held companies and do privately held companies by state if neccesary. There are more problems than just the size of the zone transfer and the load on a small number of DNS servers. There is also the problem of distributing the delegation of name space registration. This is only a problem if you are currently stuck with the job. Maybe someone from the Internic would care to comment. Curtis
The whole point of a name space is to make things easier to remember.
This is the old invariant. It's varied. I'm still waiting for a proposal for a hierarchy which will hold 25,000,000 names and still have them all easy to remember, or easy to guess from external knowledge. The new "whole point of" the DNS name space is to provide names which change less often than IP addresses change. If we could do just that we'd be winning.
Maybe .com can be split by country that you are incorporated in and in the US moved under .us and further split by state of incorporation (or state in which you do primary business for unincorporated businesses and masquerades).
An excellent idea, in my opinion.
Perhaps in the US we could also do symbol.stock-exchg.com.us with a requirement that you be listed on that stock exchange (and maybe even delagated to the exchange), covering publicly held companies and do privately held companies by state if neccesary.
Almost anything that puts depth in the tree is "good". However, we should work toward a handful of different top level strategies -- and that hand is almost full already so we can't add indefinitely.
Date: Thu, 27 Jul 1995 09:13:18 -0400 From: Curtis Villamizar <curtis@ans.net>
That a great idea. Just replace the dots in your numeric address with underscore and make it part of the domain name. :-) :-) Then you have to remember that cnidr is in reston and what their numeric address is.
I hope people recognize friendly sarcasm. No offense intended Bill. The whole point of a name space is to make things easier to remember.
Exactly! And there are few schemes as unfriendly as long string of numbers. But that does not mean that the idea is not a good one. Adding a 2 digit random number to the top or second layer of the DNS hierarchy is less than ideal, but mnemonic research has shown that short numbers (actually, up to 4 digits, if I remember the research correctly) are easily remembered, especially when grouped with other familair values. How many people on this list were ever at Paul Vixie's house when he actually was in SF.CA.US? But the mnemonic keys of familiar abbreviations ("SF", "CA", and "US") are still easy for people to remember. Even if people don't know the meaning of the symbols or if they are meaningless, as long as they are short, they are memorable. How many people know what ANS means? I'll bet not nearly as many as know who curtis@ans.net is (in a network, if not personal sense). Adding 2 or 3 random digits to .com would not significantly impact memorability of names any more than old postal zones did. (You must be at least 30 and probably near 40 to remember these ancestors of US ZIP codes.) People would quickly come to "know" that IBM was ibm.49.com just as millions of Americans knew the postal zone of Spiegel, Chicago 9, Ill. John Romkey was one of the first to truly understand the future scope of the network with his "Romkey toaster" comment. About 6 year ago I sat in meetings in DC to discuss the future of Internet directory services and could not get most of the attendees to begin to understand the magnitude of the scaling problem the Internet would someday see. The meeting included many very bright people and the final report (RFC1107) called for planning for directory service to eventually support 10 million users in up to 100,000 organizations in the US. It was explicitly assumed that only the "science and research community" needed be accounted for. My suggestions that these numbers were at least an order of magnitude too small did not get serious attention. But today I would suggest I was also too conservative. The world is a big place and naming on a global basis is a not easy. I believe there will be some ugliness and the sooner we start working on the problem, the sooner we may have a workable, if not wonderful solution. R. Kevin Oberman Energy Sciences Network (ESnet) National Energy Research Supercomputer Center (NERSC) Lawrence Livermore National Laboratory (LLNL) EMAIL: koberman@llnl.gov Phone: +1 510 422-6955
Michael Dillon writes:-
This would seem to indicate that geographical domains are a bad idea and that domain names should be based on some characteristic that is less likely to change over time.
The issue is surely one of who is prepared to do the work, and not one of routing or co-location. The present 3-letter domains have little in common wrt geography (eg some .com and .org sites are in Africa), but rather they have a lot to do with the association between the organisations. If you have a city-based domain structure, then you have the problem of the address changing (or should change, of course in practice this will be resisted) when someone moves to a new area code within the USA, just like a telephone number changes (which of course it does). Because this problem exists with telephones does not mean that it has to exist on the Internet. I'd like to think that if someone left South Africa, or someone from the US spent some time out here, that the text form (ie what humans remember) of their address would not have to change (and I don't mean simply the forwarding of email via a .forward file). If some is willing to register a hacker.us domain and run it properly, IMHO life would be a lot easier than requiring hackers to change their address each time that they move. Mike -- Mike Lawrie <mlawrie@apies.frd.ac.za> Manager: Uninet Ph: +27 12 841 3542 Foundation for Research Development Fx: +27 12 804 2679 P O Box 2600, Pretoria 0001, RSA ^^ +27 12 349-1179 sometime in 1995
If you have a city-based domain structure, then you have the problem of the address changing (or should change [...]
For some reason that I have never understood, whenever the idea of geographical domain names is mentioned, many people assume a restrictive scheme in which folk will be forced (or at least encouraged) to rename if they move. There is a spectrum of geographical domain naming schemes. At the restrictive end of the spectrum are schemes that force you to register a name associated with the area where you live or do business, and force you to relinquish your old domain name and register a new one if you move to a different geographical area. At the permissive of the spectrum are schemes that permit you to register domain names that are associated with geographical areas that you have never even visited, much less lived in. I happen to think that a fairly permissive geographical scheme makes sense. Let folk register theirname.country if they have a big enough presence in that country to satisfy whatever requirements are imposed by the country domain administrator (national companies and large universities would probably qualify). Let them register theirname.province.country if they have a big enough presence in that province/state to satisfy the requirements of the provincial/state domain administrator (medium sized companies and most universities would probably qualify). Otherwise let them register theirname.district.province.country or theirname.city.province.country if they live or have a business presence in the relevant district/county/town/city. Let the city authorities create a suburb.city.province.country level if they want to. Let sufficiently large, well-known or important geographical areas register higher up in the hierarchy than might otherwise be appropriate (for example, suburb.province.country instead of suburb.city.province.country, or city.country instead of city.province.country). Run all these in parallel, so that folk can choose between theirname.city.province.country and theirname.county.province.country if both domains are appropriate. Permit folk to keep their domain name if they move out of the area in which it was registered. Permit them to register in districts/cities where they don't live, at the discretion of the district domain administrator (for example, allowing folk who don't live in Hollywood to register theirname.hollywood.ca.us might lead to scaling problems, but allowing folk who don't live in Izotsha to register theirname.izotsha.kzn.za is much less likely to lead to scaling problems). --apb (Alan Barrett)
participants (15)
-
Alan Barrett -
bmanning＠ISI.EDU -
Curtis Villamizar -
Geert Jan de Groot -
George Herbert -
Gordon Cook -
Henry Clark -
Interpath Hostmaster -
Kevin Oberman -
Michael Dillon -
mlawrie＠apies.frd.ac.za -
Nicolas Williams -
Nikos Mouat -
Paul A Vixie -
Vadim Antonov