I would expect predictability to be useful in a diffserv context
(it will make it easier to configure classifiers). 

It might also make multihoming easier to think about.

   Brian

Erik Nordmark wrote:
> 
> > My recollection of the discussion is that some people think that
> > predictability is a good think and other think that randomness is good.
> > This text (maybe not very well formulated) intend to say
> > "If you think that predictability is not a good think, order randomly the
> > addresses,
> >   if you think predictability is a good thing, then here is a way to
> > achieve it"
> 
> But those are different implementors having different opinions, and it
> is up to the operators to deal with any issues relating to this.
> 
> So I think we need to understand what the operational concerns, if any,
> are in this space.
> Do they prefer predictability across nodes i.e. all nodes in the site
> makes the same choice (assuming same config), that a single node always
> make the same choice, or that a single node randomly chooses.
> I can see operational arguments for either one, but I'm not an operator.
> 
>   Erik

--------------------------------------------------------------------
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