[ppml] Policy to help the little guys

[David Williamson]

On Wed, Mar 19, 2008 at 01:03:31PM -0400, William Herrin wrote:
> The main problem with PI assignments is the systemic cost of
> announcing a BGP prefix ( http://bill.herrin.us/network/bgpcost.html
> ). If you look at the things from a different perspective, the problem
> is the lack of mechanisms by which the 27000 active AS's can recover
> the BGP cost directly from the folks announcing a prefix.

Can you explain to me how a PI /24 would be in any way different from
an otherwise identical PA /24?  If an end-site is multi-homed with a /24,
it seems to me that the cost is the same for PA or PI.  I can get a PA
/24 for multihoming pretty easily.  Why can't I get PI?

Your analysis points out the cost of announcing *any* route.  A route
is a route is a route, regardless of the arbitrary distinction between
PI and PA.  I think what you really want is to not see multihomed end
sites...that would really save costs in the DFZ, but is entirely
impractical for many organizations.  For those orgs that have a
legitimate need for multihoming, why force them to either be
contractually obligated to a provider or, alternatively, exaggerate
their needs to ARIN to get a /22?

PI isn't the problem for DFZ costs, deaggregation is.  The latter,
however, is going to be of increasing interest as space gets tight and
the entwork continues to grow.  If nothing else, it allows more
efficient allocations to be made.

-David