Re: Topological Motivation for draft-ohta-root-servers-01.txt?

[Geoff Huston <gih@telstra.net>]

>	It is the end customers.  They don't care if the root servers
>	are in or outside the country they are currently in provide they
>	get a reasonable level of service.  Neither should this draft.
>
>	The only places in the world where the end customer will be saying
>	"I want a root server in the country because there is not one" are
>	those countries where only way to get reasonable service levels is
>	to have one in the country.  I've heard plenty of time "We need
>	another root name server in Europe" but very seldom "We need
>	another root name server in <pick a European country>".
>
>	I say we need servers in Australia because I know that there is no
>	way to provide the service levels needed without physically
>	locating the servers in Australia.  But Australia is large and it
>	does not have close neighbours.


I find it impossible to agree that one of 13 root nameservers should be
located on an isloated island in the South Pacific with a population
of a few fractions of a percent of the total world because one or two
of the local inhabitants of that island feel that their needs usurp the
legitimate needs of many hundreds of millions of others.

I say we need root nameservers to make the DNS work efficiently for
_everyone_. Given that statement we need to examine what is appropriate
to make the technology work for all, and not a "lets pick my favourite
country" nonsensical trash.


   Geoff

(some countries, like Australia, are not even comprehensively internally
connected, and the transit across the street sometimes heads both around
the globe and far out to space. Countries do NOT host root nameservers,
its a network topology, reliability, and traffic optimisation thing.)