Francis Dupont wrote:
In your previous mail you wrote:
> Omitting DAD altogether removes the ability to detect and correct
> address collisions, whereas optimizations such as Optimistic DAD
> mean that while there may be a short term disruption the problem
> will be detected and corrected.
> > => in the real world this kind of problem cannot be corrected...
> The only thing you can do is to avoid to reproduce the same error again.
That's a rather broad statement to make,
since there may not be L2 address conflicts, and
EUI based v6 addresses may not be used.
=> L2 address conflicts are very uncommon. Personally I never got one.
The usual problem is a config which is copied but not updated,
and the auto-conf does not save you because it is not used for every nodes.
What really matters then is the effect on (the real)
address owner's and configuring node's applications.
=> usually both the real owner and the offender are dead. Murphy's law
makes the real owner an important server.
This is not going to happen in Nick's optimistic DAD draft.
The procedures don't affect existing neighbour cache entries (such as would be in place for an important server). Additionally, the procedure cleans up cache entries for itself, and provides for stifling of upper layer (non-)reachability indications from the tentative address.
Maybe you're thinking about another (prior) proposal.
I think it's worth getting implementors' experience
with address conflicts with DAD, without DAD and with
optimizations.
=> I am an implementor and from time to time a network manager so
I can say that:
- address duplications happen
- at least once DAD saved us from real trouble even the IPv6 network
was very small at this time
So I really believe in the DAD usefulness and if you'd like to remove
or "optimize" DAD on one of my networks my answer will be "NO!".
Well I'm not sure if optimize has any sinister meanings.
An optimization doesn't necessarily remove or undermine an existing procedure (if done correctly).
Regards
[EMAIL PROTECTED]
> PS: optimistic DAD is like optimistic russian roulette: look at in the
> chamber after to check it is empty.
That's highly emotive talk which doesn't help the discussion.
=> the optimistic DAD is an old topic now and I believe some of us don't
(didn't? :-) understand our opinion about it.
Roulette aside, I'm sure that if you can shoot a hole in the draft's technical merits, it will go away.
Greg
-------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
