Re: identifier uniqueness

[Masataka Ohta <mohta@necom830.hpcl.titech.ac.jp>]

Ran;

> On Friday, Nov 1, 2002, at 19:00 America/Montreal, Tony Hain wrote:
> > But since there is no way to ensure that an identifier is globally
> > unique, the only way to accomplish the goal is to couple them to
> > describe 'this identifier at this location'. If the location is not
> > stable, the static description is not stable.
> 
> Current identifiers are not globally unique either.  www.<foo>.com
> where it is implemented as a set of servers behind a load balancer
> is a trivial counter-example.

The point of the load balancer is that, viewed from the other end
of the communication, the entity corresponding to www.<foo>.com
behaves as if it is globally unique, 100% of the time.

A single entity consistently maintains TCP/HTTP state that the other
end assumes that the ID is globally unique.

> So I submit that we don't need "globally unique" identifiers".  We do 
> need
> "probabilistically unique" identifiers with a sufficiently high (<< 1) 
> probability.

It is not a probability issue.

							Masataka Ohta