On 23 Jun 2005, at 15:53, Mark Smith wrote:
On Thu, 23 Jun 2005 10:46:44 -0400
Vlad Yasevich <[EMAIL PROTECTED]> wrote:
I've always considered policy from 3484 an example that will be
modified by the end-user. That's why it's there. It gives a really
basic default. If you want to be multiaddresses with ULA and global
prefixes, you need to tweak things.
But for sure there are basic policies that could (should?) be enabled
by default so that ULAs aren't preferred as source addresses for
anything other than potential ULA destinations (e.g. due to sites
misbehaving by putting ULAs in global DNS RRs)
I'm rapidly coming to the same conclusion as Fred :-/
The use of site locals (sorry - i'll pay penance later) and, after
FEC0's deprecation, ULAs that we have motivation for is disconnected
networks that have topology - principally sensor nets and such beasts
as community wireless networks. Both of these application domains
have a requirement for provider independent addressing or survivable,
swift and automated renumbering that requires little to no
administrator intervention. The latter hurts my head, even after
working on an IPv6 Renumbering project ;-)
I'd think ULAs and globals deployed in parallel would be the most
common
case, so I'd think it should just work "out of the box" without having
to tweak anything.
Absolutely.
Without putting a bit of effort into it i.e. off the
top of my head, I can't think of scenarios where ULAs wouldn't be
deployed in parallel with globals.
Either scenario above can start out as just ULA with potentially
ephemeral and variable availability of globals.
Mark/
--
Dr Mark K. Thompson
Electronics and Computer Science
a School of the University of Southampton, UK
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
