[arin-ppml] Policy Proposal: Dedicated IPv4 block to facilitate IPv6 deployment

[Owen DeLong]

On Jun 8, 2008, at 7:04 AM, Leo Bicknell wrote:

> In a message written on Sun, Jun 08, 2008 at 01:35:22AM +0000, Paul  
> Vixie wrote:
>> it seems possible that a lot of people think that others are doing  
>> prefix
>> length filtering, but that only a few people, or nobody, is  
>> actually doing
>> it.  the policy process should be informed somehow.  any idea how  
>> we can
>
> First we need to think about the different cases:
>
> 1) How many networks prefix filter customers...
>     strictly based on their allocation?
>     allowing between their allocation and a /24?
>     allowing anything inside their allocation?
>
This is a constantly moving target and getting any accurate measure
of any of those three numbers would be a challenge even for the fine
folks at CAIDA, I suspect.

Anybody feel like asking KC if she can tackle this?

> 2) How many networks accept longer routes from customers than they  
> will
>   pass along to their peers?
>
My personal suspicion: Many.

> 2) How many networks prefix filter peers...
>     strictly based on their allocation?
>     allowing between their allocation and a /24?
>     allowing anything inside their allocation?
>     allowing based on RIR published minimum allocation sizes?
>
Probably more than filter their customers.

I also suspect that this number will drop in the short term after IANA
freepool runout, then, start climbing dramatically on a somewhat
parallel trajectory to the routing table growth that will result.

>
> For instance, if you believed 99% of the networks filtered on RIR
> minimum published size, it's likely if the RIR's gave out /28's
> they would allow it.  However, if you believe 99% of the networks
> filter at a /24 boundry, that would not be the case.  The generic
> question "do you prefix list filter" does not provide enough
> information to make informed decisions.

My rough guess is that currently less than 2% of networks filter on
prefix size at all.  However, I believe that at least one very large  
provider
does so (or at least did so until very recent history).

Further, I believe that the largest networks will be the ones with the
most aggressive prefix filters in the shorter term as the tables start
to grow.

Owen