[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [GLOBAL-V6] PI space under IPv6
On Thu, Jan 18, 2007 at 06:58:37AM -0800, Peter Sherbin wrote:
> > Why, exactly? The distinction is real, and points to
> > real and looming problems with the (IPv6) routing system.
>
> PI is just a requirement from users. The infrustructure to
> support it needs fixing, hence there is this discussion.
Yes, and as you know, there are ongoing discussions in
many other other fora as well . The problem here, as
I see it, is that given the user requirement (as you
point out), the RIRs have no other technology (or
whatever) available to implement it. That one is on
us. Note that I'm not saying that many in our community
haven't tried (on the contrary), but as this thread
points out, little (read: nothing) has happened since the
deployment of CIDR.
> Actually people asking for PI are more concerned with a pool of
> IDs to name/control/monitor their items/things and
> less so with routing. So split ID/locator. Give IDs to
> people. Build routing on locators. ISPs will follow the new
> architecture. Opposing ISPs will be naturally replaced by new
> willing ones.
If only it were that simple. History seems to be telling
us otherwise (see Paul's note in this thread). That
doesn't mean we should give up (I'm the poster child for
not giving up), but lets also not trivialize how hard
this has been, both in terms of the technology needed,
but also in terms of all (all) of the supporting
"infrastructure" (e.g., IETF, RIRs, SP business models,
...).
That being said, I'm heartened that so many in this
community are still willing to take a whack at fixing
this, especially given the checkered history we're
dealing with.
--dmm
Attachment:
pgp8d0isQ8LW2.pgp
Description: PGP signature
