Re: [DNA] RE: Review of draft-krishnan-dna-simple-01.txt

[Bernard_Aboba@hotmail.com]

>> [BA] The way I think of it is that the host needs to determine
>> whether its existing addresses are valid.  It can determine that based
>> on receipt of an NS from a router, or receipt of RAs.
>
> From your remark, I assume that a host verifies its addresses one bye
> one. The verification of one address doesn't guarantee the validity of
> others, right?

The host verifies a subset of routers which generated valid addresses by 
doing
NUD to them and sending an RS.  Verifying reachability via NUD confirms all
prefixes formerly advertised by that router, so it can validate
multiple addresses at once.

> This also entails added complexity. Previously DNA checks whether the
> host still remains in the same link or not, then accordingly verify
> (or invalidate) all its addresses at the same time.

Simple DNA can do the same thing, by verifying all addresses obtained
from the same router simultaneously.  Since there are typically only
a few valid addresses, it is not a complex operation to send a few NSes
and one RS.

> IMO, the host can verify address B based on the verification of
> address A. If address A and B were valid in the same link and address
> A is proven still valid, it's reasonable to assume that address B also
> is still valid (except in some corner cases.)

If router B is no longer present, and does not answer NS or send
RAs for prefix B, then the host probably should not conclude that
address B is still valid, even if router A is present and sends NA or RA.

> I have in mind the following example.
>
> Assume a host is attached to a link
> with two access router R1 and R2
> which advertise prefix 1 and prefix 2 respectively.
>
> The host autoconfigures address 1 with prefix 1.
>
> After a while, the host executes DNA and performs NUD with R1 and R2.
>
> When a solicited NA from R2 arrives to confirm R2's reachability,
> IMO, it's reasonable to assume that
> address 1 with prefix 1 is still valid,
> even though a solicited NA from R1 didn't arrive yet (or never).

Simple DNA will perform NUD to R1 and R2 simultaneous with sending an
RS.  When the solicited NA arrives from R1 it will autoconfigure address 1.
If R2 answers the NS, it will autoconfigure address 2.   It will also 
process
the RAs from R1 and/or R2.  However, if R2 doesn't answer the unicast NS
or the RS, then it won't auto-configure address 2 until it hears evidence 
that
R2 is still there.