Re: [v6ops] PI [ULA draft revision #2 Regarding isolated networks]

[Mark Andrews <[e-mail]>]

In message <[e-mail]>, Gert Doering writes:
> Hi,
> 
> On Mon, Jun 02, 2014 at 12:58:33PM +1000, Mark Andrews wrote:
> > There is also nothing to prevent a stack trying multiple source
> > addresses when it hasn't been explicitly bound to a source address
> > by the application.  
> 
> Well, yes, nothing *prevents* this.  But given lack of IETF guidance on
> this, stack implementors do not seem to be keenly interested in actually
> implementing this.
> 
> (Does BIND do this...?)

We don't automatically try multiple source addresses per family but
you can specify different sources addresses for different destination
addresses.

At this point multiple PA prefixes are not common yet.

ULA + PA (or PI) just works at the moment.

That said I would most probable implement it something like <D1,S1>,
<D2,S2>, <D3,S1>, <D1,S2>, <D2,S1>, <D3,S2> assume S1 and S2 are from
different PA prefixes.

On top of detecting network failures nameservers also have to deal
with broken nameservers / firewalls that drop queries because they
don't like something in the query despite it being perfectly in
spec.

Then there is RRL processing which drops queries if the server sees
excesive traffic as part of amplication mitigation.

> Gert Doering
>         -- NetMaster
> -- 
> have you enabled IPv6 on something today...?
> 
> SpaceNet AG                        Vorstand: Sebastian v. Bomhard
> Joseph-Dollinger-Bogen 14          Aufsichtsratsvors.: A. Grundner-Culemann
> D-80807 Muenchen                   HRB: 136055 (AG Muenchen)
> Tel: [phone]           USt-IdNr.: DE813185279
-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: [phone]                 INTERNET: [e-mail]