Static vs SLAAC - Static expected to be preferred?

[Mark Smith]

On Wed, 27 Apr 2011 11:22:09 -0700
Doug Barton <dougb at dougbarton.us> wrote:

> On 04/27/2011 11:20, sthaug at nethelp.no wrote:
> >>> I disagree with that, as when I specifically choose to go to the effort
> >>> of overriding an automated configuration mechanism with a static one,
> >>> then I intend for and expect the static configuration to be used in
> >>> preference to the automated configuration.
> >>
> >> I agree.  RHEL5 appears to prefer a SLAAC address over a static for outbound
> >> connections, which was definitely a surprise to me.  I remember thinking
> >> "why would I want *that*?"
> >
> > Fully agreed here too. If I have configured a static IPv6 address I
> > expect it to be used in preference to a SLAAC address.
> 
> I'm confused. If you're assigning static addresses, why are you also 
> using SLAAC?
> 

Any time you have a mix of hosts where you'd want some to have statics
(e.g. servers, printers etc.), while others just use SLAAC (desktops,
laptops. Think Internet cafes, small offices with a single LAN, and
probably less commonly home networks. This would be functionally
equivalent to how IPv4 is comonly run today, with static addresses used
for servers, and DHCP used for assigning addresses to desktops, laptops 
etc.

The original example was a bit uncommon - the VPS service provider
wanted customers to have SLAAC addresses for monitoring, while
customers wanted and were assigned ranges of static addresses, all on
the same link and with both static and SLAAC addresses coming from
within the same /64.

So Doug, in those sorts of scenarios, if you configured a server with a
static address (using the default preferred and valid lifetimes of
infinite), would you expect that static address to be used in
preference to any other addresses the server acquired (with lower
value preferred lifetimes), such as SLAAC addresses?

Thanks,
Mark.