[arin-ppml] Rationale for /22

[William Herrin]

On Tue, Jul 28, 2009 at 1:46 PM, Kevin Kargel<kkargel at polartel.com> wrote:
>> AFAIK, there are no ISPs allowing customers to multihome with less
>> than a /24 currently but many who are allowing those /24s.  So if that
>> Org gets a /24 from ARIN or a /24 from it's ISP, there is the same 1
>> entry in the table.
>
> This is incorrect.  If the Org gets a /24 from it's upstream ISP there is
> the same entry in the routing table.  If the Org gets a discontiguous /24
> from ARIN there is an additional entry in the routing table.

Kevin,

For the multihomed case we're discussing, this statement is in error.
In order to aggregate small routes into a larger route, two things
must be true:

1. The covered address space must be contiguous in a netmasky way
2. The network topologies for the two routes must be the same. That
is, the entrances from "the Internet" must be the same for both
routes.

In the multivendor multihomed case, each of the customers ISPs
connects to the rest of the Internet differently. The network topology
for the multihomed customer generally differs from the network
topology of any of the ISPs serving the customer. As a result, his
route is not aggregable.

In the single homed or single-vendor multihomed case, the customer
shares a network topology with the ISP. As a result, his route can be
shared with the ISP's route as long as the addresses are contiguous.



On Tue, Jul 28, 2009 at 11:52 AM, David Williamson<dlw+arin at tellme.com> wrote:
> While I think mast of us are concerned about routing table growth, it's
> a problem that is unrelated to good stewardship of the address space.

Hold your horses there David. As I believe Tony Li over on the IRTG
RRG said, addressing is routing is addressing.  The two are
inseparable. If they weren't then maximum conservation of IPv4
addresses would obviously come from having the IANA assign every /32.

> That's why routability of any given block is explicitly note to not be
> guaranteed by the NRPM.

Quantum physicists don't guarantee that oxygen molecules won't
randomly leave your vicinity until you suffocate. Would you suggest
that ARIN's non-guarantee has more substance to it than the
physicists'?

Let's be practical here: guarantee or no, we intend that addresses
assigned by ARIN be routable on the Internet. And we endeavor to meet
the routing system's needs so that they will be.


On Tue, Jul 28, 2009 at 2:09 PM, Jon Lewis<jlewis at lewis.org> wrote:
> The distinction some people may not be getting is that if I know ARIN
> allocates from a /8 nothing longer than /20s, then if I'm running out of
> routing slots, I can use a prefix-list to ignore anything /21 (or maybe /22)
> or longer from that /8.  If ARIN allocates /24s from a /8 or probably longer
> net, then I need to accept those /24s.  That's the theory anyway.

Hi Jon,

That theory was destroyed by NRPM 4.2.3.6. If you filter on RIR
minimums you will damage your connectivity to small multihomed sites.

Regards,
Bill Herrin


-- 
William D. Herrin ................ herrin at dirtside.com  bill at herrin.us
3005 Crane Dr. ...................... Web: <http://bill.herrin.us/>
Falls Church, VA 22042-3004