There's an article today (.Mobi's Case For Mobile Domains) that makes me nervous.
If you read this blog regularly, you know that I believe that new top level domains should be granted to pretty much anyone who can demonstrate technical competence and a willingness to abide by relevant internet technical standards. This is really a corollary of what I call "The First Law of the Internet".
I believe that .mobi, or rather the technology approach that it represents, could contain a significant "public detriment" by virtue of a potential lack of adherence to the spirt of some internet standards.
The article mentions that .mobi is needed because mobile devices change addresses and that due to the caching in DNS, DNS data will tend to be out of data as machines move. This is not a new concern; we had the same concern back in 1987 when we did RFC1001/RFC1002.
(I do not know why the .mobi folks think they need a new TLD to solve this problem - they could just as easily do it under a second or third level domain. Nor do I understand why they want to turn DNS on its head and use it as a rapid-change distributed database. It seems to me that they could get a lot more bang for their efforts by establishing rendezvous servers and using something like LDAP to query and update those servers.)
What scares me is the fact that the article indicates that mobile devices will tend to lose their DNS information when they are power cycled. This means that every time one of these things is powered up it is going to have to go to its DNS hints, re-learn the root zone name servers, re-learn the major TLD name servers, etc.
DNS only works (or to use proper internet terminology - DNS only "scales") because of caching. Without caching the traffic loads on the root servers and TLD servers (and other DNS servers) would increase many times over.
The article indicates a potential naivete the part of the mobile industry with respect to DNS. Given the number of mobile devices (cell phones, pagers, etc) out there today, a little naivete could mean a big risk that core DNS servers could be overloaded with queries that could have been obviated through the use of a bit of "clue".
Posted by karl at March 23, 2004 12:58 PM