Michel Py wrote:
Michel, maybe my mind isn't lateral enough, but I
can't think of an example of anybody who would want
to pay for guaranteed globally unique site local
addresses. Usually people seem to be happy with
"good enough and free" verses "perfect and not-free".
Are there any examples you might be able to give ?
Any organization that is serious about using site-locals. Two examples
comes to mind:
- Corporate mergers (avoids the renumbering pain found today when the
two corporations were using 10.0.0.0)
- VPN connections.
Ah, yes. The corporate merger case... at the MERGER time they are
desperate and willing to spend money, but it is already too late.
We need a solution that is good enough and free for the time when
the network addressing gets DECIDED. Often, this is done when the
companies were small start-ups and they have a single-address uplink
and some network of devices internally.
My thinking is that Pekka's statistically unique site local approach
is the right solution here. Also, let's take a look at the requirements
posted by Margaret:
> - persistent addresses to put in configuration files,
> security filters, etc.
Given a MAC/vendor id/company name string -based approach we
can do this with Pekka's approach.
> - addresses that can be used in intermittently connected
> sites and remain valid
CAn do this too.
> - addresses that survive ISP renumbering
Yes.
>Without the use of ambiguous addresses.
Yes.
> - work properly when two organizations merge
Yes.
> - be possible to associate with an administrative
> entity to which they are registered
I tend to think this is not a requirement. Might even be
harmful. But yes, this is doable as well. Say, we could
REQUIRE that the address is a hash of the MAC or the
company name. So at least you could show that your address
is yours, though it doesn't prevent others from claiming that
as well (given the small number of bits). Optional registry
is doable as well, but like I said above that may not be
desireable.
Finally, I'd like to add a requirement: We need addresses that
a group of nodes can come up all by themselves without the
participation of humans. Pekka's addresses can do this too.
Jari
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
