[arin-ppml] 2010-8: Rework of IPv6 assignment criteria

[David Farmer]

On 9/15/10 13:06 CDT, Owen DeLong wrote:
>>
>>> The HD-Ratio is replaced with a simplified 75% utilization threshold
>>> based on nibble boundaries for end-user assignments.   This threshold is
>>> somewhat more restrictive for larger assignments, while slightly less
>>> restrictive for the smaller /44 assignments, than the HD-Ratio.
>>> However, in both cases it is much easier for an end-user to understand
>>> the policy criteria that applies to them.
>>
>> This means that a different type of measure would be applied to allocations made to ISPs and assignments made to end users.

Yes, it would be a different measure.  However, there are number of 
significant differences in how ISPs and end-users make use of addresses 
anyway. So I'm not sure a different measure is inappropriate.

> Presumably, the reason for the difference is not that end user organizations are not capable of understanding the HD-ratio concept. And anyway, if it is difficult to understand, ARIN can be asked to produce explanatory materials.

I want to refer people to RFC 3194: "The Host-Density Ratio for Address 
Assignment Efficiency: An update on the H ratio" and section 6.7 of the 
NRPM.

https://www.arin.net/policy/nrpm.html#six7

Actually, while I believe many people do not understand HD-Ratio, 
especially end-users, that isn't the only issue. if it were, then 
"explanatory materials" could maybe solve the problem.

As I see it, the HD-Ratio of /56 prefixes as applied in ARIN policy is a 
reasonable measure for ISPs and LIRs as they make assignments to 
end-users.   However, I don't believe that the HD-Ratio of /56 prefixes, 
is a meaningful measure to apply to an end-users.  With that measure, 
how many /64 subnets is an end-user suppose to consume out of a /56 
before they can get another one?  How about the same question for a /48? 
If they use one /64 from 184 /56s, are they entitled to an additional 
/48?  That is one way I can read the policy, but I know most people 
think that you should have to use more than 184 /64 subnets to get an 
additional /48.

>> If the concept underlying the HD-ratio is fair, is it fair not to use that concept when calculating the amount of space available to an end-user organization?

Yes, it would be possible to write the policy using HD-Ratio of sties as 
the threshold rather than a 75% threshold.  However the threshold for 
/44 would have been 14 sites instead of 12, so for that case I liked 75% 
better.

This is the 75% threshold that is in the current Draft Policy;
More than 1 but less than or equal to 12 sites justified, receives a /44 
assignment;
More than 12 but less than or equal to 192 /sites justified, receives a 
/40 assignment;
More than 192 but less than or equal to 3,072 sites justified, receives 
a /36 assignment;
More than 3,072 sites justified, receives a /32 assignment or larger.

This would be the equilivant using HD-Ratio of /48 sites;
More than 1 but less than or equal to 14 sites justified, receives a /44 
assignment;
More than 14 but less than or equal to 184 /sites justified, receives a 
/40 assignment;
More than 184 but less than or equal to 2,487 sites justified, receives 
a /36 assignment;
More than 2,487 sites justified, receives a /32 assignment or larger.

I contemplated using, the 75% threshold for /44 and HD-Ratio for the 
rest, but it seemed even more complicated to explain the reasoning 
behind that idea.  So, after thinking about it for a while, I decided to 
just go with the 75% threshold across the board.  The combination of /48 
per site and assigning on nibble boundaries, seems to more than makeup 
for the fact that the 75% threshold is a little more restrictive than 
HD-Ratio.  And, yes the 75% threshold has the added benefit that it is 
much easier for most people to understand.

But I have provided the numbers above so everyone can judge for themselves.

> Leo,
>
> HD-Ratio is intended to take into account the losses due to hierarchy inherent in scaling provider networks.
>
> End user networks tend to be flatter (within a given site) and thus not subject to those losses.
>
> As such, we felt that a simpler, easier to understand guideline was more appropriate to the circumstance.

While all end-user networks are not necessarily flatter, I agree the 
majority are.  However, I think the /48 per site and assigning on nibble 
boundaries will serve even large hierarchical end-user networks as well 
as HD-Ratio in most cases.

Furthermore, very large hierarchical end-user networks, especially for a 
large multinational corporation, are not precluded from considering 
themselves as an ISP or LIR, and then starting with a /32 and being able 
to avail themselves of HD-Ratio.  That might be more appropriate anyway, 
the networks for large multinational corporation with many independent 
business units, probably have much more in common with ISPs then the 
majority of other end-user networks.

-- 
===============================================
David Farmer               Email:farmer at umn.edu
Networking & Telecommunication Services
Office of Information Technology
University of Minnesota	
2218 University Ave SE	    Phone: 612-626-0815
Minneapolis, MN 55414-3029   Cell: 612-812-9952
===============================================