when & how could policy be changed

[Stephen Sprunk]

At 10:51 01-07-97 -0700, you wrote:
>I don't think anyone is certain what preconditions would need to exist in
>order for a policy change that would grant every bona-fide multihomed ISP a
>PI prefix but we can see the possibility of having operational realities
>considered as part of the policy since the network operators will likely
>all be members of ARIN.

I'm sure network operators around the world will be watching intently at
the first ARIN meeting; this issue is sure to come up.  What are the
expectations on proposal/voting procedure for ARIN?  Can we expect a policy
decision during/after the first meeting?

>But I don't think we can clearly see how routing table sizes, dampening
>algorithms and filtering will work into the equation until we have some
>substantive discussions among ISPs.

I doubt the large ISPs will take a position for or against this proposal
prior to the first ARIN meeting.

>Right now a lot of them are either not on this mailing list or are keeping
>quiet for some reason.

Unfortunately these days most people "agree" by not saying anything.

>I personally would like to see some PI space opened up with longer prefixes
>than /19. This could be a new /8 like 210/8 that we all agree to allocate
>in /20 blocks. Or we could use reclaimed space from the swamp and allocate
>it in /20 and/or /21 sizes. In the case of 210/8 we need providers to agree
>to adjust their filters.

Out of curiosity, what would be the rationalization for using more of the
(almost depleted) class C networks instead of a class A?  There's close to
a hundred A's available, and RIPE-155 has shown that a class A is perfect
for this kind of plan.

Is anyone interested in debating the conditions, size, or class for such
allocations?

>But before we can decide just how this should be done we need some hard
>numbers, especially on how many additional routes the new PI space would
>add.

In the case of 4096x /20 PI allocations, it would be reasonable that within
the first 2 months the number of routes will increase by 4096 (and hit the
50k mark again).

The trick, however, is to allocate them and place restrictions on them such
that after 60 days an equivalent or larger number of routes would be
dropped, as small ISPs were able to transition out of the PA blocks they
currently advertise as more-specifics.  I'm sure the large ISPs would be
happy to increase filtering on the PA blocks to "encourage" the return of
said more-specifics.

>And we also need some more thorough analysis of the prefixes that appear
>to be eligible for aggregation in the weekly CIDR reports.

It'd be nice if there were some way to penalize ISPs on the "most wanted"
list.  Perhaps denial of new allocations based on measurable routing
inefficiency?


Stephen