Re: [savi] A few comments on draft-vogt-savi-framework

[Eric Levy-Abegnoli <elevyabe at cisco.com>]

Christian,

Christian Vogt a écrit :

On Oct 28, 2009, marcelo bagnulo braun wrote:

  

For this case, we need a prioritization of which address assignment
methods take precedence over which other address assignment methods.
And you are right that the definition of this prioritization will
require care.

      

yes, moreover, we need to figure if it is enough with a prioritazation
or we need to change the mechanisms in order to interact properly.  
As i
stated earlier (and eric noted before me), we certainly need to update
send savi to properly interact with the slaac savi (at least, haven't
doen the analysis for dhcp savi yet)
    

Correct; the prioritization may affect the protocol specifications
(SLAAC, DHCP, SEND) just as well as it affects the framework document.

My suggestion -- as a working group member, not as a chair -- is that
the framework document should talk about the prioritization
conceptually, whereas the protocol specifications need to include the
details that implement the prioritization.  Thoughts?

  

I don't see much value in a document that describes things at the "conceptual" level without looking at the details of address assignament methods.
Its going to be very confusing if each "protocol" document describes the prioritisation of binding entries from it's own point of view.
So far, I see s 1)SLACC 2)DHCP 3)CGA 4)MANUAL (static binding) 5) LOCAL (L3 addresses configured on the L2 device). And then there are combinations. For instance CGA is relatively orthogonal to the protocol (SLACC, MANUAL, DHCP, etc...).
I thought the very purpose of the framenwork document was exactly to describe those interactions, which include prioritzation, but also protocol interactions. This draft addresses none of this. Does that mean we need a document that describe the interactions?  That could be draft-levy-abegnoli-savi-plbt-01. I had given up on it, hoping the framework document would cover the same topic. So far, it does not.
Eric

- Christian


_______________________________________________
savi mailing list
savi at ietf.org
https://www.ietf.org/mailman/listinfo/savi