Re: new draft

[Iljitsch van Beijnum <iljitsch@muada.com>]

On zondag, mei 11, 2003, at 18:30 Europe/Amsterdam, Michael H. Lambert 
wrote:

> A couple of comments on your draft:
> 1)  If I'm drawing the correct inference, it would appear that the 
> number of locator addresses used by a client goes as n!, where n is 
> the number of ISP prefixes.  I can easily see sites with n=5.
A multihomed server would use n locator addresses for a single 
identifier.

A multihomed client could decide to create double prefix addresses for 
each combination of prefixes, which would make for n(n-1) addresses, or 
even double this by also creating two versions of each combination, 
with different prefixes being the identifier for each of those two 
versions. I'm not sure if I'd recommend this, though...

> 2)  In the case of a host with multiple interfaces, is each interface 
> treated separately (both in the DNS and wrt locator addresses)?
I don't think it matters whether n addresses are from n interfaces with 
one address each or 1 interface with n addresses. So it seems this 
solves host multihoming as well.

Iljitsch