[arin-ppml] Rationale for /22

Owen DeLong owen at delong.com
Fri Jul 31 15:11:09 EDT 2009


On Jul 30, 2009, at 8:30 PM, Leo Bicknell wrote:

> In a message written on Thu, Jul 30, 2009 at 07:17:20PM -0500, David  
> Farmer wrote:
>> You are assuming that the number of multi-homed users will remain a
>> constant, if you allow a multi-home user with a PA based /24 from a  
>> provider
>> to switch to a PI based /24 from ARIN.  It is entirely possible  
>> that the
>> availability of PI /24s from ARIN for multi-homing could create an  
>> inducement
>> for more users to multi-home, just so they can get PI space.
>
> David summed it up well, however, a few notes.
>
> If a network multihomes with a subnet of a PA block and someone
> somewhere else in the routing table filters that route but not the
> PA block, there is still full reachability.  If someone multihomes
> with a PI block and that same person filters the same route, there
> is no reachability.
>
While this is sort of true, the reality is that the point of multihoming
is to have connectivity even if one of your providers goes away.
Given that, if the route is filtered, the functionality of multihoming
is partially lost or degraded.

> Thus the size of the set of all routes is the same; but the size
> of the filtered routes but still global reachability set is reduced.
> There are a number of people trying to use the second set.
>
Not really, but, sort of kind of enough that some people choose
to buy into this myth.

> While the most traveled road is Multihome with PA, then Multihome
> with PI, there are plenty of folks who single home with PA (no
> global routing table slot) until they can qualify for an ARIN PI
> block, then they multi-home.  There are users who go from nothing
> to multi-homed; for instance startups who make sure they qualify
> day 1 for a multi-homed prefix.  Presumably if the barrier was lower
> more would do so.
>
Sure, but, I'm not buying that these numbers are large enough for
that to be a bad thing.

I would agree with policy limits that prevent people from getting
multiple /24s rather than trading up until they reach /22 or possibly
even /21.

> I suspect many of the folks on this list would spend a lot more
> effort trying to multi-home if they could get 1 /32 from ARIN for
> $10 per year.  The number of cable modem and DSL customers asking
> their providers for BGP would skyrocket.  It's not because those
> folks are multi-homed with PA space today.
>
I think that /32 is reductio ad absurdum.  However, /24 is not and
I think that there is an under-served class of users still stuck with
a relatively arbitrary and in some ways punitive policy that we
should have rectified long ago.

Owen




More information about the ARIN-PPML mailing list