Re: [DNA] Simple DNA [Fwd: I-D Action:draft-krishnan-dna-simple-03.txt]

[Bernard Aboba <Bernard_Aboba@hotmail.com>]

> Receiving an RA confirming a prefix should be OK too. Do we have to 
> handle this (manual addresses) as a specific case? I was thinking of 
> just including text in the document specifying adding an entry into the 
> DNA address table for this manually assigned address along with the 
> associated prefix and the router from which we heard the prefix. 
> Everything else should work fine after this.

Putting the manual address into the DNA address table is fine.  
With respect to Simple DNA, I think that handling of manual addresses
is actually simpler than it was with DNAv4. 

In working on DNAv4, there was a concern about the 
implications for probe retransmission.  Normally a unicast ARP 
probe might be sent only once or twice.  For a dynamic address 
this was not such a big deal, since other mechanisms (e.g. DHCPv4) 
were running simultaneously and they would retransmit; as a result,
the host would be likely to get an address as long as the DHCP server
was up, even if the ARP probes were dropped for some reason. 
However, for the manual address case, without probe retransmission,
failure rates for DNAv4 could be quite high.   For example, consider a
wireless network with high frame loss and only a single probe being
sent.  Failure rates of 10% or more could be experienced. 

With Simple DNA, I think the situation is better because even if the
NS probes are lost, the RS will still be periodically retransmitted.  So
overall, the probability of a false negative is lower.