[AusNOG] IPv4

Terry Manderson terry at terrym.net
Mon Mar 4 17:19:37 EST 2013

Hi Mark,

On 04/03/2013, at 12:21 AM, Van Der Meulen, Mark wrote:
> Whilst everyone else gets on the “I’m thinking about everyone in the rest of the world where they are so underprivileged, and I’m also helping your grandmother get IPv6 on her iPad, blah blah” in attempts to passive aggressively big note how great they are without sounding self centered, I would like to add my perspective. The perspective of the white male in his mid twenties who grew up in an affluent country and who is for the most part, quite self centered…. (since apparently that matters when discussing IPv4 resource policy now)

I tend to believe that folks are really drawing on the sense that without everyone doing the *same thing*, with respect to the protocol level, nothing works. Harking back to the old days - it really is a case of which flavour of UUCP do you run? And while it seems self centered, it is generally from an organisational perspective. Remember the volume of participants of this list is based around networks that earn revenue not only from having just their network behave as expected, but also the rest of the internet 'work'. 

> I’ll start off by saying that really the whole “ you think you have it bad, you should see everyone else in the third world” argument is simply evasive behavior. Also, whilst you’re pushing how selfless you all are in helping others, think on this:  If you want to take credit for being the persons who have been trying to make the internet a better place through your policies regarding IPv4 and IPv6, then you also need to take the credit when flack is handed out about your inability to properly handle IPv4 resource management, and how the bad decisions that you may have been part of are affecting others.

To be honest, this is a fact of life - what is seen now as a 'good policy' may well end up turning into mud. I co-authored and presented the policy that allows everyone to get IPv6 easily, by default if they have IPv4 already. I think at this stage its a "good thing". However as time goes on there probably will be a point that I hold my hat in hand and profusely apologise for such a error in judgement. 

> From this point on in my email, I wish for the “IPv6 is here and IPv4 is long gone” argument to remain irrelevant and for those of you who are defending APNIC to try look at things a little more objectively. There are apparently(according to this rather long email thread) ~16,384 /22’s left, and to some people that’s a lot of IPv4 left – not only that, there is a lot of unused/unjustified/unneeded space out there so let’s talk about that for now.
> I work with a number of corporates in banking and finance, and in addition to this a number of small ISP’s and hosting companies trying to get by. Most of the small/medium sized companies I work with are providing IPv6 in a dual stack, and for the most part are limited to their /22 or less, which they have received in the past 1-3 years. On the other hand working with the larger companies, I see absolutely no intention whatsoever of moving to IPv6. They purely run IPv4 environments and show blatant disregard for the amount of IPv4 resources they consume through their various legal entities. Generally, this is because of a number of reasons:

This is one area that I think deserves recognition. In terms of engaging the 'corporate' stakeholder in the APAC region the policy side has missed the mark. I don't actually think there is a blame point - honestly this is uber-geek network operations and not in the core business of banks, retailers, legal firms, and others like that. (however I do note that they are taking an interest, and you are evidence of that)

It could be said that more is needed. More awareness should be promulgated. More beating of drums. More something..

> a)      They simply don’t care. They have far more address space than they need, and they have “better things” to do with their time than manage a migration to IPv6.
> b)      They don’t understand IPv6, and are concerned about its security implications.
> c)       Management would never approve a project to IPv6 as it simply isn’t in their interest commercially.
> d)      It is a pain when dealing with compliance.
> e)      Why?? Everything works right now, so why make life harder for ourselves? We are never going to use all of our 
> allocations.

Equally don't underestimate the agility of commercial corporates. While IPv4 may be going the way of the fax, the minute the lack of v6 affects a customer relationship, a share price, a profitability line, you will see management approve projects to make *stuff* happen.

So while www.coles.com.au or www.woolworths.com.au doesn't push v6 now, if the situation arose that they might loose market share by _not_ running v6 or some protocol translation - I would happily bet my WOW and WES shares that they would move heaven and earth to ensure they remained current.

> So on one hand I see, large corporate A with more address space than required, abusing the system because they can’t be bothered(amongst other reasons) and it doesn’t make commercially sense to. Then I see small company B implementing IPv6 but is severely limited ini its growth abilities because of the lack of IPv4 available to it – why is this? Because if they are selling to consumers, the consumer wants a working, no fuss connection that works with everything and with minimal fuss. If they don’t get the minimal fuss, they can and will take their business elsewhere in a twinkling of an eye. If they are selling to businesses, then there is a good chance that the business customers have a very similar attitude to that of company A, in which case they only want IPv4 services. So you get the picture, even though they have IPv6, no one wants it because they don’t care enough about it, and you can’t grow a business off the back of a product that no one wants.
> My point here is that companies who are abusing the system are actually making themselves more commercially viable – they are lowering or maintaining capital and operating expenditure and for the most part making sound commercial decisions. What this means materially for the companies not in the same position(quite likely because of timing) is that they are forced to adopt a strategy which can potentially make them less commercially viable because they are the ones which need to increase CAPEX and OPEX, without any real prospect of those investments returning profit or increased customer retention.
> Let me make this very clear, the very same policies that have allowed the company which has obtained excessive address space through a registries mismanagement to be more commercially viable are the very same policies that are limiting the commercial viability of another company. I would say that by definition, this makes the policies of the registrars anti-competitive in nature.


Unfortunately what you describe is the side effect of any scarce resource and the interests of the stakeholders setting the policy, and even to such an extant as how a policy is created through the policy development process. (food for thought: with all this technology we still seem to need a policy proposal to be presented at a APNIC meeting to reach some level of in-room consensus before moving on [http://www.apnic.net/community/policy/process]) 

Sadly, I see that has parallels with the oil and gas trade associations. The internet policy frame, while allowing more engagement from stakeholders still carries a quorum (albeit self selecting) of folks within this particular industry space.

> Is it the problem of the small/medium/large business that the address space was previously mismanaged? Is it to something to do with their margins or their own lack of planning? Absolutely not, they are generally making an effort at their own expense to implement other solutions, when often times companies who are abusing the system clearly are not attempting to implement other solutions, and the fact that a company may or may not have the margins to buy more space on the open market is largely irrelevant, they need to be given an equal opportunity to grow - just like a company buying new IPv4 resources or a corporate who already owns lots of IPv4 resources is given or has been given that opportunity.

So here you say "abusing", I don't see it as that. I think this is just plain ignorance, and simply - you don't know, what you don't know (Rumsfeld?, I think?)

And really why would they know? their focus often isn't on this internet 'tool' but on their core business.

> At the end of the day APNIC and the other registrars were around long before many of these companies started doing business and long before many of these companies even knew what an IP Address was and how it affected their business model. They have been responsible for the allocation of address space for quite some time and hence these registrars must take full responsibility for their lack of judgment and lack of ability to properly administer allocations – they are

So it's worthwhile to note (and maybe others have done this in other posts) that the organisations known as the RIRs are not supposed to set policy. They are there to enact the policies that the community (us?) put forth via the policy development process. In fact the RIRs are only, by their own charter, allowed to do things as directed by the policy development process and their membership.

> being paid to have the very foresight that they have proved they never had(and I’m not talking about 20 years ago, I’m talking about in the last 5 years). The new/middle aged companies trying grow are simply casualties of a poorly designed and executed set of policies, IPv6 won’t fix it for them now and they likely won’t be viable long enough for it to fix it for them later. Higher margins/CGNAT/etc are not solutions that will provide long term fixes, and are not solutions commercially competitive enough to find a place in a business plan.

You could argue that the entire RIR framework is wrong, and it could have been done differently - and I (and many others) would love to spend many hours drinking ale and planning how it _could_ have been better/faster/stronger. However even in a new and different regime of address allocation, we would still have been limited to 2^32 addresses. With all the other protocol constraints that exist to maintain the end-to-end principle I just don't see how that is ultimately going to make make much of a difference. So even if I came up with a super-duper approach, I hazard a guess that I could have gotten maybe 10 more years out of IPv4. But still back where we are now.

How we move on from here now, is key. 

> This is my point, as an outside observer of how the APNIC policy influences the Australian market that I work in, I have good reason to believe that the APNIC policies are cultivating anti-competitive environments. Due to the very nature of the beast in that those who govern it are also those who benefit from it, I don’t believe it will change or can change.

Actually, I would like to be idealistic enough that if the policies are, as you say, truly anti-competitive then they will change, or at least evolve. Not to would open the system up to other pressures.

> Full Disclosure: I very recently applied for IP space through APNIC for one of the smaller companies I work with who will very likely run out of space within 6 months of having an allocation, so I am biased(quite clearly, given the above email!)
> Full Disclosure: I run IPv6 at home and everywhere I can, so please don’t try discredit me because you found someone I work with who doesn’t run IPv6 – it’s poor form and proved you have no interest in correcting bad policies.
> Full Disclosure: My grammar is pretty average, even for an engineer. Sorry if this hurt your eyes.
> Lastly, I’m young and feel empowered by voicing my opinion on matters I have only have half the picture on. So ignore me as you wish.

Actually it's been an interesting read.


More information about the AusNOG mailing list