[arin-ppml] early experimenter /32 catch-22

[Christopher Morrow]

On Fri, Oct 8, 2010 at 10:39 AM, Jason Schiller <schiller at uu.net> wrote:
> On Fri, 8 Oct 2010, Christopher Morrow wrote:
>
> |On Fri, Oct 8, 2010 at 9:38 AM, David Williamson <dlw+arin at tellme.com> wrote:
> |> On Thu, Oct 07, 2010 at 04:08:37PM -0400, Christopher Morrow wrote:
> |>
> |>> (you can't guarantee longer prefixes than a /32 get global
> |>> reachability... so subnetting the current allocation isn't feasible)
> |>
> |> Is that really true?  Would someone fess up about current routing
> |
> |I know of a case (solved by announcing the /32 aggregate) of this, yes.
> |('hello docs.python.org!')
> |
> |> practice?  We definitely accept longer prefixes, as we only have a /45,
> |> and we only advertise /48s.  Of course, that's in the PI block.
> |
> |sure, so in the PI ranges it seems most places (save vzb?) accept to
> |/48. In the other allocation sets you can get to /32 but not often
> |very much further (for full reachability).
>
> AS701 accepts and sends PI and PA /48s to Peers for about 8 months now.

awesome, thanks for accepting my bribe. (note on the list)

> Seems most (not all) folks I talk with accept PI /48s, and slightly better
> than half accpet PA /48s.  Renysis says it PA /48 reachability is much
> worse than my experience.  (See the NANOG prior to the previous)

yup, it seems at least in some adhoc looking that PA /48's are not
always reachable without a covering route.

-chris