North American Network Operators Group|
Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical
Re: Statements against new.net?
On Tue, 13 March 2001, "Brett Frankenberger" wrote: > > On Tue, 13 March 2001, "Steven M. Bellovin" wrote: > > > Put simply, deploying multiple public DNS roots would raise a very > > > strong possibility that users of different ISPs who click on the same > > > link on a web page could end up at different destinations, against > > > the will of the web page designers. > > > > Its not really the "will of the web page designers." If this becomes > > popular, I suspect most web page designers will start using dotted-quad > > addresses inside their HTML URLs on their web pages. So clicking on a > > link on a web page will go to were the web page designer directs you. > > Except for NAT strangeness, IP Address are mostly globally unique. > > Fine. Then we have a technical issue associated with "all the links > break when a web server is moved to a different IP address because > whoever was hosting it had to renumber", and it's companion "I can fix > the links because there's no way to know what the new IP address is". > > Since the reason renumbering is a reality are technical -- or, are at > least presented that way -- the need for a single DNS root remains > technical. Generally the web page author is nominally in charge of moving a web page from one IP server address to a different IP server address. If this happens, they could change (perhaps through a script) all the links on the page to use the new IP address just like they currently renumber all the other things which must change when moving. In general, people clicking on URL's have no idea what the anchor URL is. Unless you look at the HTML source, you see the text not the URL. Of all the protocols which use DNS, HTTP/HTML seems to be the easiest for users to use something else. TELNET, FTP, and SMTP have much more user-visible DNS names, and are therefor much more difficult to change. There are several inter-related issues (as always): - Long-term stability of the identifier - Global uniqueness (within the relevant domain) - Easy for humans to remember - Simple for machines to process