[ppml] Technical reason why /52,/56,/60,/64 are bad

Edward Lewis Ed.Lewis at neustar.biz
Tue Aug 21 11:20:38 EDT 2007

At 16:21 -1000 8/19/07, Randy Bush wrote:
>>  The Assignment Guidelines proposal attempts to dictate what size
>>  prefixes LIRs (ISPs) assign to their customers.
>and who the hell gives arin this wonderful new prerogative to tell me
>how to run my business?  and what will you do when i completely ignore
>it because it is none of your business?

I apologize for not resisting (further) to read the entire thread 
before posting.  I want to add on something to Randy's comment.

In 2005 I saw the development and presentation of an IPv6 idea at 
ARIN (Orlando), RIPE (Stockholm), and APNIC (Hanoi).  After 
witnessing this, I realized this:

1) The RIRs only care about the assignments because they want to 
measure utilization (primarily when a request for more comes in).

2) The RIRs ought to stay away from any recommendations about operations.

3) The RIRs ought to stay away from any recommendations about the 
protocol.  (As in, how many bits are needed in the "host part.")

Given this, the proposals ought to avoid "assign" these prefixes but 
rather emphasize "we will measure utilization on these prefixes."

Violating #2 resulted in rat holing on aggregation practices and 
elicits comments like Randy's above.  (No offense Randy...)  I recall 
the "rebirth" of the class-full space in IPv6 comment at APNIC.

Violating #3 resulted in rat holing on new uses of IPv6 (addressing 
ice makers) and why so many bits are in the host part of the address.

I urge the policies to document what ARIN does - measure utilization 
in this case - and stick to just that.  If ARIN wants to measure 
utilization of private residences on /56's and sub-netted networks on 
/48, that is reasonable but don't give the impression of dictating an 
addressing strategy.
Edward Lewis                                                +1-571-434-5468

Think glocally.  Act confused.

More information about the ARIN-PPML mailing list