[ppml] Revising Centrally Assigned ULA draft
bicknell at ufp.org
Mon Jun 18 14:56:22 EDT 2007
In a message written on Mon, Jun 18, 2007 at 11:42:40AM -0700, David Conrad wrote:
> I am extrapolating based on behaviors I see being discussed or
> implemented in all the RIRs regarding the liberalization of PI
> allocation policies. Your extrapolation of my extrapolation made
> assumptions that John is pointing out aren't necessarily valid. The
> availability of PI to anyone who asks does not imply that the address
> space is routed nor does it imply PA space magically goes away or is
I agree with your statements, however I think we have to make some
educated guess at what is probable.
Regarding the second, we have a long history in IPv4 that given the
choice between PA and PI, people would rather have PI. The only
barriers that have kept people from getting PI are RIR policies and
cost. So I would imagine if the bar to get PI were set "as low as
possible" that most (80%?) of the people with PA today would move
Now, once we're in that situation, let's look at routing. First,
keep in mind we're in a situation now where the lions share of the
allocated prefixes are /48's. The tool used today to limit the
number of routing entries taken up is prefix length. Well, if I
route so much as a single one of the PI /48's, and I want to use
that tool I have to allow them all.
Can I generate a prefix-list that contains everything in the DFZ?
Probably not. Do I trust my Internet Neighbors to only allow in
prefixes that "justify" a routing slot? Heck no, someone is going
to charge money to get the slot and let anyone get it, and then
everyone else will have to do the same or be at a competitive
disadvantage. Does crypto solve the problem? No, same issue, the
provider will sign it for a fee.
We need a method of deciding "worth" that can be implemented
independently by each ISP and does not require them to trust their
neighbor. Today that is prefix length. I've seen nothing in any
policy proposal or in any of the active IETF groups that seems to
have even a remotely workable solution. Thus, IPv6 is falling back
to the same thing as IPv4, prefix length, and more "important"
people get larger blocks.
Leo Bicknell - bicknell at ufp.org - CCIE 3440
PGP keys at http://www.ufp.org/~bicknell/
Read TMBG List - tmbg-list-request at tmbg.org, www.tmbg.org
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 187 bytes
Desc: not available
More information about the ARIN-PPML