Forcible reclamation?

Paul Ferguson pferguso at CISCO.COM
Tue Jul 8 19:09:59 EDT 1997


At 05:49 PM 07/08/97 -0500, Stephen Sprunk wrote:

>The point was that allocating PA IPs to a multihomed AS is pointless; it'd
>be simpler and easier for everyone involved to just give them PI space.
>Aggregating between the AS level and the continental level is basically
>impossible; aggregating continent(s) might be possible, but isn't currently
>effective.
>

Gee. Thanks for the incomprehensible, and incorrect, overview.

Aggregation at the AS boundary is exactly what we do now, kimosabe.

Aggregation on geographical boundaries is not possible. We've
discussed on multiple occasions, and on multiple mailing lists,
over the course of the past few years.

I would contend that technically, it works just fine. What you
are clamoring over is a human-induced situation -- a few large
providers filtering smaller announcements. Changing this policy
is where you should be focusing your efforts, not changing the
allocation policies.


>There is no "issue of increasing the address space"; the same number of IPs
>will be used, it's just a question of whether the multihomed AS gets them
>as PI from the NIC or PA from their provider(s).  It's obvious which would
>be more stable for the user (PI), and which would have a higher likelihood
>of mismanagement (PA).
>

You've obviously not comprehended my earlier missives.

The argument thus far from participants of this list for IPv6
has been in response to statements that since IPv4 address space
is a finite, limited resource, and cannot be arbitrarily allocated,
a migration to a larger address space [IPv6] is a desired goal.

I have contended that proposing a migration to v6 based solely
on this reasoning this is flawed logic -- allocation policy
would still be needed with IPv6. And chances are, you guys
would still feel discriminated against when you encountered
situations where a larger provider filtered smaller, more
specific prefixes, which violated aggregation of larger
aggregated announcements.

The current proposals on IPv6 address structure also have
provisions for aggregation -- at least one proposal has
two levels (a top level aggregator and a 'mid'-level
aggregator). The same issues will exit within IPv6
which exist within IPv4, and that is entrenched
efforts to keep the global routing system stable
and functioning.

Comprende?

- paul




More information about the Naipr mailing list