Excerpts from Olivier Bonaventure on Wed, Jan 21, 2009 09:52:12PM +0100:
> > I think the work on EID allocation guidelines for RIRs is premature at
> > this stage.
> 
> I proposed this work item because I think that we need to integrate the
> EID and RLOC allocation mechanisms in the development of the protocol
> itself. I don't think that "doing as we've always done with IP
> addresses" is the best solution.

Olivier: I certainly hope I misunderstand you.  Choosing a particular
forwarding or mapping protocol shouldn't lock us into a particular
allocation mechanism.  <shudder>

> For example, I consider that mapping systems must be secured from day
> one. This will probably require the utilisation of certificates (e.g.
> similar to the X509 certificates discussed within the SIDR WG) to bind
> EID blocks to owners. These certificates could then be used to secure
> the mapping replies.
> 
> The "EID allocation guidelines" that I mentionned are more related to
> this security issue than to the size of the blocks to be allocated and
> other issues.

OK now I understand a little better.  I would not want to build an
assumption about a particular security mechanism into the protocol,
even if it did allow us to optimize it a lot.  We can optimize
ourselves into an evolutionary dead end.  Consider MD5.  It's better
to have at least one mechanism that works satisfactorily and can be
replaced by a better mechanism in the future.

Scott
_______________________________________________
Int-area mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/int-area

Reply via email to