[arin-ppml] Draft Policy ARIN-2014-17: Change Utilization Requirements from last-allocation to total-aggregate - revised

Owen DeLong owen at delong.com
Thu Jul 17 21:23:19 EDT 2014


On Jul 17, 2014, at 17:42 , Michael Peddemors <michael at linuxmagic.com> wrote:

> On 14-07-17 05:13 PM, Owen DeLong wrote:
>> Asked and answered earlier in the thread, but I will give it a shot...
>> 
>> If you have a collection of smaller blocks, it is much harder to get to 80% utilization in each one while still preserving contiguous free space to accommodate a larger-than-normal customer or a surge in customers. For example, if you have 3 /22s and have utilized all but a /26, three discontiguous /27s, and a two discontiguous /28s and a couple of /29s of the last one and all but a /29 of each of the other two and you have a customer come in who needs a /24, you're stuck. You need to find some other customer with a smaller need or some valid utilization for a small block or two in order to reach your 80% utilization on that last block.
>> 
>> OTOH, if you look at your overall situation, you're well past 80% overall.
>> 
>> This is not an uncommon scenario (in various forms) for small providers. The rules as they exist were written with minimum allocations of /20 and larger in mind at the time and smaller organizations mostly did not receive space directly from ARIN.
> 
> Thanks Owen,
> 
> And I guess maybe that helps define where you and me differ on this, this again seems to represent the scenario of a 'hoster' who ends up 'renting' out his IP(s) to other parties.

How is that different from a rural residential provider who ends up selling to a combination of residential and business subscribers or even mostly business subscribers?

Why is a hoster an invalid use of IP addresses? Colo and Datacenter operators have been ISPs in this region for a very long time now and I've worked for a few of them.

It's at least part of what my current employer does.

> I have to make an assumption that this is a lot rarer than you think, those entities get larger space.

I have to say that you are very wrong there. Yes, there are entities that you are more likely to know that get larger space, but for every large entity that you know, there are literally hundreds of much smaller entities simply trying to get by in smaller markets, targeted market segments, and other vertical integration scenarios. The internet is a VERY diverse place and the small players outnumber the large ones by many orders of magnitude.

> And where I have concern is for the small service operators who need that /24/23/22/21/20 for their own operations or business models or cloud services, but don't want to be beholding to any provider and want to 'own' their own IP Space.

Those are "end users" in ARIN parlance and I don't think they would suffer under this policy. In fact, they would be largely unaffected by it. The only extent to which it might effect them is if it accelerates runout. I don't believe for a second that this proposal will represent a significant acceleration for runout beyond the satisfaction of long pent-up demand for addresses that has gone as yet unsatisfied due to policy quirks.

I suppose if you want to protect some small to medium end-users at the expense of small ISPs and you consider that fair policy, there is little I can do to convince you otherwise. It seems quite unfair to me. I believe that small ISPs have just as much right to the address space as end users of any size.

> These are the ones that I think may suffer here, and they don't usually have a voice here.. and I am being a bit altruistic possibly, but I find it hard to support ideas which make things easier for those who want the 'real estate' value vs the end users of IP space.

I think that's a very jaded way to look at it and I don't consider this helping those that want the "real estate" value. Are you saying ARIN shouldn't issue space to ISPs at all and that end-users should get absolute priority over ISPs in general? If not, I am really having a hard time understanding the argument you are making here.

> End users you can be sure will fill all the 'nooks and crannies' of their total space.  And while your test case may not seem bad, when you look at those with /16/15/14/13../8 Those 'nooks and crannies' become large swathes that should be used up before they get more as we run low.

Sure, but when you consider the three month maximum that those providers are facing, they won't get enough more to avoid filling up the nooks and crannies anyway, so I'm not thinking it really has as much of an effect as you seem to think.

> The current policy helps to ensure that all the delegated space gets used first.. before giving an incumbant more .. again IMHO

No, it actually doesn't. What it does, instead, is reward those who can renumber on a nimble basis and shove stuff into their most recent allocation while leaving larger holes in their earlier allocations.

The proposal would actually require that ALL space is utilized to at least 80% while the current policy only requires a strict 80% on the most recently used block.

> My two cents..
> 
> I want to see more full utilization before running out, (in the absense of ways and methods of 'stewardship' and priority on the types and validity of usage.) as a better chance that people in the future can still get some from ARIN for longer.

I want to see more IPv6 deployment before running out. In reality, we'll both likely be disappointed in different ways.

Owen




More information about the ARIN-PPML mailing list