[Manet-dt] RE: SMF - New Field Source ID

["Joe Macker" <[e-mail]>]

Ian/ALL;
Thanks for your note.

First the ID update plans and whats happening with the editor. I'd like some
quick DT feedback on several items but plan to have something ready before
the ID deadline.  This could be ready next week.

I began editing the document several weeks ago by removing all previous
SMF-specific NHDP design.  We have agreed as a group to target MANET NHDP.
This is pretty much done and is the major change.  Most implementations I
know of presently use an existing MANET unicast protocol and borrow the CDS
information so implementations are still functional without NHDP.    

Also Pedro and Brian Adamson contributed some gateway interface thoughts and
draft write-ups.  Particularly multiple gateways design approaches and
issues.  After reviewing and thinking about that I am inclined to keep the
gateway section to an operational issues discussion.

This is still some disagreement on how to best solve particular gateway
issues.  Those deisgn particulars and practices may be best described in a
separate document.  I am still a fan of a sequence-based approach to support
efficient DPD implementation processing for high speed flows.

Regarding you SOURCE ID suggestion.  Ian I think your idea is potentially
useful especially the way you stated when included. That way we can still
support naitve IP source specific multicast flows as well without
encapsulation or header extensions.  I am ok with crufting that carefully
into the header extension section. Others?


>-----Original Message-----
>From: Ian Chakeres [mailto:[e-mail]] 
>Sent: Monday, September 18, 2006 9:38 PM
>To: [e-mail]
>Cc: Joe P Macker (home)
>Subject: SMF - New Field Source ID
>
>Currently SMF DPD is based on the source address (and often the
>destination) and the ID field. I'd like to create another 
>field called "source ID" that would take precedence over the 
>source address when included.
>
>I myself envision this used in conjunction with 
>autoconfiguration. For example, a DHCP request could flood the 
>network while the source might not have an address yet. In 
>this case, the source IP address would not be useful in 
>ensuring uniqueness. Instead a randomly generated ID could be 
>used to associate the message (or messages) with a particular 
>source and avoid unnecessary duplicate packet transmissions.
>
>What do you think?
>
>Ian Chakeres
>
>BTW: I noticed the SMF draft is no longer active. We should 
>try to get a new rev out before the IETF 67 deadline.
>



_______________________________________________
Manet-dt mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/manet-dt