> The statement I was responding to was "the idea that a prefix can be
> changed at a whim is just a fantasy.". The point is that arbitrary &
> random prefix changes are reality and believing it doesn't happen is the
> fantasy. 

Okay, let me rephrase that:

The idea that a prefix can be changed at a whim without having a serious
impact on the reliability and performance of applications is just a fantasy.

People who think that DNS is a sufficient replacement for stable IP
addresses haven't actually bothered to look at how poorly DNS works, 
how ambiguous and unstable DNS names can be in practice, how quickly 
apps could expect to recover from address changes (especially when the 
zone servers themselves have moved).  Nor have they investigated the 
burden of expecting apps to implement explicit app-level acks for 
everything that gets sent over a TCP stream (so that the app can recover 
from a connection broken by renumbering), because without such acks
the app doesn't know how much data has actually been read by its peer.

> While it is nice to believe that providers that do this are
> subject to customer retaliation, where are the customers going to go? If
> the SP otherwise provides the best service available (or increasingly
> the only service), the renumbering even is a fact of life.

I don't disagree with that - in fact I currently have to live with an
BRI ISDN connection to my house because no provider that can provide
higher bandwidth is willing to give me a stable IP address.

But just as we cannot force providers to be reasonable, neither can
we force applications to work under unreasonable conditions.  The best
we can do is to try to define what is reasonable and what is unreasonable.

Keith
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to