[ppml] Policy Proposal 2002-3: Micro-Assignments for Multihomed Networks

[william at elan.net]

On 21 Aug 2003, Jeff S Wheeler <jsw at five-elements.com> wrote:
>
> On Thu, 2003-08-21 at 15:24, McBurnett, Jim wrote:
> > I hate to see the /22 in this, as I am multi-homed and I don't foresee
> > the ability to use it in a year. But I understand the routing table issue.
> 
> I'm sure small, multi-homed shops that need provider independent space
> will simply inflate their utilization figures to demonstrate need of a
> /22, perhaps wasting a little address space in the process. I can live
> with that as /24 assignments are an overwhelmingly large portion of the
> routing table today, and a few slightly larger blocks will probably cut
> into the growth of /24s a bit.
This "honesty" coming from the person responsible for taking somebody elses 
ip block (hijacking four old micro-assignments in fact) is very interesting...
(for reference, see http://www.completewhois.com/hijacked/gang_ings.htm). 
Obviously you have no problem lying to ARIN either about ownership of ip 
block or in the future about size of ip block you may need. Unfortunetly 
policies created by ARIN are all designed based that people using them 
would be fairly honest (and if this is a problem - we can work on it). 
 
> The key thing to understand here is, when you create limitations like
> this, ISPs will fudge their numbers to meet the arbitrary minimum
> network sizes so they can get the type of assignment they need. How many
> organizations have /20s today that could do with a /22? 
Hopefully, not as many as you Jeff Wheeler thinks (do not ever assume
everyone else operates the same way you are). I can tell that all companies
I've dealt with that have /20s all had enough use of those ip blocks and 
have not exadurated numbers when getting them.

> These largely arbitrary limits, while well-intentioned, only make the 
> ARIN difficult to work with.
ARIN is already very difficult to work with (and this is topic for 
completely different discussion), the current proposal for 2002-3 will not 
change difficulty in working with ARIN. If anything on that issue I support
ARIN being difficult and doing hard checking about micro-assignment requests.

-- 
William Leibzon
Elan Networks
william at elan.net