external issues in dns scalability (1995) (was Re: Namespaces)
avg@kotovnik.com (Vadim Antonov) writes:
Just stop issuing alhpanumeric domain names and use numerals only.
In 1995, Brian Kahin invited a bunch of folks to come to DC and talk about DNS ("a policy wonk's dream"). Not understanding that the whole thing was a sham, I wrote a paper and presented it. The audience (hi Vint!) reacted in stunned silence, and probably had a good laugh about it later at the bar. I found the postscript files and converted them to PDF and put them online: http://www.vix.com/~vixie/bad-dns-paper.pdf http://www.vix.com/~vixie/bad-dns-slides.pdf All I can really say is: "I told y'all so." Vadim, thanks for reminding me that there was a time when these problems were still soluble.
-----BEGIN PGP SIGNED MESSAGE----- Paul Vixie wrote:
avg@kotovnik.com (Vadim Antonov) writes:
Just stop issuing alhpanumeric domain names and use numerals only.
In 1995, Brian Kahin invited a bunch of folks to come to DC and talk about DNS ("a policy wonk's dream"). Not understanding that the whole thing was a sham, I wrote a paper and presented it. The audience (hi Vint!) reacted in stunned silence, and probably had a good laugh about it later at the bar.
The papers bring back memories. Good and bad. Even Postel submitted a draft proposal to use .0 to .9 and .a to .z. Not a bad idea.
All I can really say is: "I told y'all so." Vadim, thanks for reminding me that there was a time when these problems were still soluble.
Wait, wait. Some of the operational ideas in your paper have been implemented! The root servers were split from the TLD servers. There are multiple registrars. How close are we to having signed root delegations? -----BEGIN PGP SIGNATURE----- Version: PGPfreeware 7.0.3 for non-commercial use <http://www.pgp.com> iQCVAwUBOqe3P9m/qMj6R+sxAQGEsAP9H39AvVMiclveldXJHa0K7v4m9J3B9hHf 5j+FRQPnnm9EV1nazSOi58OnbUHorb7N146qItRsVsOeUt+ag+oftZnqSyO9QC6p TX9visFX2DjOO1RmOQhYzOp4xIjs2Tf+aoULrrkH0BEGiWzgtSOdTYBkaT5SRjVd hJvFnELcQNc= =8TiD -----END PGP SIGNATURE-----
[ On , March 8, 2001 at 07:43:37 (-0800), Paul Vixie wrote: ]
Subject: external issues in dns scalability (1995) (was Re: Namespaces)
All I can really say is: "I told y'all so." Vadim, thanks for reminding me that there was a time when these problems were still soluble.
Wow. Re-reading that paper now is like deja vu all over again! Seems like it was many decades ago.... I think, IIRC, I agreed with many points you made back then (and especially the controversial ones like, avoidance of government (except maybe the UN), closing .COM et al, the support of ISO country code domains, etc.), and I'm almost certain that I fully agree with all of them now. Even though we sort of have shared-registry types of things happening now I'm not so sure it would hurt to go back to one registry per gTLD. The problem is that it all hinges on closing down .COM, et al.... -- Greg A. Woods +1 416 218-0098 VE3TCP <gwoods@acm.org> <robohack!woods> Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>
On 8 Mar 2001, Paul Vixie wrote:
I found the postscript files and converted them to PDF and put them online:
http://www.vix.com/~vixie/bad-dns-paper.pdf http://www.vix.com/~vixie/bad-dns-slides.pdf
All I can really say is: "I told y'all so." Vadim, thanks for reminding me that there was a time when these problems were still soluble.
Thanks, Paul, the paper is excellent. Reminds me why i established the rigid geographically-administraive domain allocation scheme for organizational domains in .SU back in 90 :) This was unfortunately abandoned by the .RU registry folks, with the resulting .COM-like chaos. Anyway, my current position is no matter what you do, people will complain and find ways to subvert the system. When big money moves in there's absolutely no hope to protect registries (and their policies) from threat of lawsuits or other forms of intimidation and political pressure. Thus, any unique naming scheme is unworkable, no matter how you organize it. The only way the humanity found so far to ensure uniqueness of any kind of names is to back naming regulations with the coercive power of states. Therefore, the choice is either to remove the cause of the problem (i.e. the intrinsic value of names, resulting from their human-readability), or to get national laws passed in order to create enforceable name allocation policies. Whatever Next Generation Internet is going to be there, i would argue that it should not include DNS at all. Otherwise it will eventually be a subject to licensing-style regulation by states (i.e. domain allocation). I am personally is not a big fan of state involvement - because the very next thing to happen after states taking the power to allocate a critical resource (i.e. names) would be the call by various pro-censorship groups to deny allocation of names to any kind of sites they consider objectionable. (Hmmm... local and state governments in US are known to go as far as to rename creeks and towns whose names were disliked for some reasons by some vocal groups :) On a more philosophical note - the existence of any centralized essential resource historically always guaranteed that the control of the entire system depending on the resource unfallingly passes to whoever manages to seize that resource, usually to the government. There is no reason for the Internet to have such vulnerability (and most governments out there are not decent, by any measure). --vadim
participants (4)
-
Paul Vixie -
Vadim Antonov -
William Allen Simpson -
woods＠weird.com