------- Blind-Carbon-Copy X-Mailer: exmh version 2.2 06/23/2000 with version: MH 6.8.3 #1[UCI] From: "Steven M. Bellovin" <smb@research.att.com> To: Michael Sondow <msondow@iciiu.org> cc: JIM FLEMING <jfleming@anet.com>, DOMAIN-POLICY@lists.netsol.com, Karl Auerbach <karl@cavebear.com> Subject: Re: The PSO on TLDs... Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 02 Nov 2000 14:47:06 -0500 Sender: smb@smb.research.att.com In message <3A018E13.D6EA0113@iciiu.org>, Michael Sondow writes: >JIM FLEMING wrote: >> >> Steve Bellovin wrote: >> > >> > 1) We see no technical problems caused by the creation of a >> > relatively small number of new gTLDs. >> > 2) We recommend that the root servers be monitored for increased >> > load caused by these new gTLDs. >> > 3) We suggest that they be semantically "far" from existing gTLDs, >> > to avoid confusion. Thus, .corp would be be bad, because it is >> > too similar to .com. > >> By small number...can we assume 2,048 would be small ? >> >> Given that the .COM servers used to be co-resident with the root >> of the IPv4 legacy DNS, it can easily be shown that supporting >> millions of .COM names was not that different from supporting millions >> of TLDs from a traffic and protocol point of view. Since you claim to >> be providing "PSO" input, I assume protocols come into play somewhere >> in these discussions. >> >> With respect to your item #2, keep in mind that many ISPs do not >> rely on any root servers to locate the clusters of TLD servers. With >> Floating Root technology, one can easily locate the dominant TLD >> servers for a TLD with a quick bottom-up check. This reduces the >> reliance of an ISP on the single point of failure of any RSC and allows >> the ISPs to filter out errors that pop up in the various RSCs. You'll probably have to forward this to DOMAIN-POLICY@LISTS.NETSOL.COM, since as I recall it only permits subscribers to post, and I'm not on that list. > >Rather than opinions based on technical "protocol" issues, Mr. >Bellovin's comments seem to be politically inspired. The gist of his >message is: few new TLDs, very carefully chosen. > >One wonders if his post to Louis Touton was spontaneous, or if it >was a response to a call from Touton for support from the PSO for >his (Touton's) TLD policies. The ICANN board requested the opinion of the Protocol Council on the general question of creating new TLDs. The points quoted above were the consensus of the PC. There was no hint whatsoever of a desired answer. > >Who, by the way, is Steve Bellovin? Who else is on the Protocol >Council? How were they chosen? Are any of them not employees of >mega-companies, like AT&T, which have been instrumental in the >creation of ICANN (i.e. the GIP combine)? > Who am I? There's a quote from "Lord of the Rings" that I'm fond of. ``Who are you, Master?'' he asked. ``Eh, what?'' said Tom sitting up, and his eyes glinting in the gloom. ``Don't you know my name yet? That's the only answer. Tell me, who are you, alone, yourself and nameless.'' If you want more details than that, feel free to peruse my Web page: http://www.research.att.com/~smb. The Protocol Council is comprised of two appointees from each of the four Protocol Support Organizations: the IETF, W3C, ITU, and ETSI. You can find other details at http://www.pso.icann.org/, including the current membership. I will add only that each supporting organization uses its own procedures to appoint its representatives; for the IETF, the IAB made the appointments, in keeping with the IAB's role as liason to external organizations. I'm unaware of any sinister plots, by AT&T or anyone else, behind my appointment. I assure you that AT&T does not control the IAB. And my service on the IAB -- as with all other participation by anyone in any IETF activity -- is as an individual, rather than as a corporate representative. I certainly did not solicit any input from anyone else in AT&T when we were discussing this question, nor did I receive any. As for the other questions -- the notion of "root server clusters" is dealt with in http://www.iab.org/iab/IAB-Technical-Comment.txt. I personally don't know how many new TLDs would be too many; I do note that (a) we have only one root, which therefore must be managed much more conservatively, (b) we are indeed seeing signs of stress on the root servers, so adding too much more load to them would seem to be unwise, and (c) prudent engineering practice suggests that before making major changes, we try a pilot program first and see what happens. --Steve Bellovin ------- End of Blind-Carbon-Copy =========================================================================