Re: [Roll] Last Call Review of draft-ietf-roll-efficient-npdao

[Rahul Arvind Jadhav <[e-mail]>]

Thanks Eric for the review and feedback.
Regarding compatibility with old deployed nodes, we have received similar feedback (from Shwetha Bhandari) before. May be it is better to make this explicit in the document. Please check if the statement below makes sense.

NEW
                This document assumes that all the 6LRs in the network support this document. If there are 6LRs en-route DCO messaging which do not support this document, then the route invalidation for corresponding targets may not work or may work partially i.e., only part of the path supporting DCO may be invalidated.
Alternatively, a node could generate an NPDAO if it does not receive a DCO with itself as target within a specified time limit. The specified time limit is deployment specific and depends upon the maximum depth of the network and per hop latency. Note that sending NPDAO and DCO for the same operation would not result in unwanted side-effects as detailed in Section 4.6.2.
END

For other comments, find my responses inline.

Thanks,
Rahul

Issues:

In the introduction, I have some trouble parsing “an optional messaging” – perhaps it was meant to just be “optional messaging?”

[RJ] Yes this has to be fixed. Will remove “an”. This has been brought to notice by others also!

There are difference between the abstract (which claims the document describes issues with NPDAO) and the Introduction (which goes further to say that the document also discusses requirements and specifies a new message intended to meet the requirements and solve the problems.

So, it’s a problem statement, requirements document and solution specification all rolled into one.  I guess this is not a problem for anyone in the WG?

The title for section 1.3 should probably be “Why is NPDAO Important.”
[RJ] I am not sure if question mark here is incorrect. This section title is supposed to indeed read as a question.

In that section, I have to wonder if saving memory is the most important factor in removing route information that is no longer valid.  What about issues with forwarding packets toward destinations that can no longer be reached along that path?
[RJ] Our initial motivation to work on this problem was memory issue. But yes, I agree that in P2P traffic scenario the forwarding will be a problem. This has been described in Section 3.3

Perhaps I misunderstand the approach, but it seems to me that the problems mentioned in section 2.1 should self-resolve.  The fact that one node cannot send messages to another node to inform that node that it has no DAO information seems kind of irrelevant if the node, or the link connecting it, is no longer there.

Perhaps the issue you’re really trying to describe is with unreliable links and nodes, as opposed to missing links and nodes?  That would make sense in a low-power and lossy network.

If the behavior described in section 2.2 is common, and a more reasonable behavior is not anticipated by the original NPDAO specification, than I agree that is a problem.

If the problems described in 2.1 and 2.3 do exist, it seems likely that the underlying issue is really about an assumption of reliable delivery where that might not be the reality.

[RJ] That's true. Unlike NPDAO, DCO depends on “more” reliable links thus alleviating (but may not completely eliminate) the problem of route invalidation.

All of this said, the 3 requirements seem reasonable at least as far as they go.  There seems to be an implied requirement to introduce at least some level of reliability (as I seems intended by providing the DCO-ACK). And there probably should be a requirement to support compatibility with deployed implementations.

AFAICT, this document does not describe how nodes using the newly specified messaging and behavior would be compatible with deployed nodes.  Minimally the document should be clear if the intention is not to provide for compatibility.  There are at least a couple of indications that there is deployment.

[RJ] Yes. Have specified the corresponding text above. Kindly check if it makes sense.