On Mon, Oct 5, 2009 at 11:39 PM, Doug Barton <dougb@dougbarton.us> wrote:

As a practical matter we're "stuck" with /64 as the smallest possible network we can reliably assign. A /60 contains 16 /64s, which personally I think is more than enough for a residential customer, even taking a "long view" into consideration. The last time I looked into this there were several ISPs in Japan who were assigning /60s to their residential users with good success. OTOH, a /56 contains 256 /64s, which is way WAY more than enough for a residential user. The idea that a residential user needs a full /48 (65,536 /64s) is absurd. OTOH, assigning a /48 to even a fairly large commercial customer is perfectly reasonable. This would give them 256 /56 networks (which would in turn have 256 /64 networks) which should be plenty to handle the problems of multiple campuses with multiple subnets, etc.

Keep in mind that not all 'fairly large enterprises' are willing to sit at a single ISP, they may have diverse offices on diverse network provider connections. They may want the easy of saying: 'All my address blocks are in 1.2.0.0/16' and not understand (or like) that they now have to deal with wierd routing and addressing problems because they can't get a /32 and break it up into /48's all over creation (different ISP's/links/etc) or deal with the split of address space they'd get from ISP /48 PA assignments. the enterprise world has changed quite a bit from IPNG's early days... Someone who runs a large Enterprise with global office locations and who's actually deploying ipv6 internally/externally ought to give a presentation at NANOG and/or IETF. I don't disagree with the math I snipped, I do appreciate you laying it out, and I don't think that there are a super large number of folks in the scenario I layed out above. I've seen quite a few at previous employers though... -Chris

Doug Barton wrote:

[ I normally don't say this, but please reply to the list only, thanks. ]

I've been a member of the "let's not assume the IPv6 space is infinite" school from day 1, even though I feel like I have a pretty solid grasp of the math. Others have alluded to some of the reasons why I have concerns about this, but they mostly revolve around the concepts that the address space is not actually flat (i.e., it's going to be carved up and handed out to RIRs, LIRs, companies, individuals, etc.) and that both the people making the requests and the people doing the allocations have a WIDE (pardon the pun) variety of motivations, not all of which are centered around the greater good.

I'm also concerned that the two main pillars of what I semi-jokingly refer to as the "profligate" school of IPv6 allocation actually conflict with one another (even if they both had valid major premises, which I don't think they do). On the one hand people say, "The address space is so huge, we should allocate and assign with a 50-100 year time horizon" and on the other they say, "The address space is so huge, even if we screw up 2000::/3 we have 7 more bites at the apple." I DO believe that the space is large enough to make allocation policies with a long time horizon, but relying on "trying again" if we screw up the first time has a lot of costs that are currently undefined, and should not be assumed to be trivial.

I agree with the point about undefined costs, but the biggest one that is easy to see is that 100-300 years from now when someone thinks about moving on to the second /3, this entire discussion will have been lost and there will be an embedded-for-generations expectation that the model is cast in stone for all of the /3's.

It also ignores the fact that if we reduce the pool of /3s because we do something stupid with the first one we allocate from it reduces our opportunities to do "cool things" with the other 7 that we haven't even thought of yet.

www.tndh.net/~tony/ietf/draft-hain-ipv6-geo-addr-00.txt shows a different way to allocate space, using only 1/16 the total space to achieve a /48 globally on a 6m grid. Other ideas will emerge, so you are correct that we can't assume we have 8 shots at this, but if the first pass is really bad the second will be so draconian in restrictions that you will never get to the third.

In regards to the first of the "profligate" arguments, the idea that we can do anything now that will actually have even a 25 year horizon is naively optimistic at best. It ignores the day-to-day realities of corporate mergers and acquisitions, residential customers changing residences and/or ISPs, the need for PI space, etc. IPv6 is not a "set it and forget it" tool any more than IPv4 is because a lot of the same realities apply to it.

Well mostly. It is not set & forget, but a lot of the day-to-day in IPv4 is wrapped up in managing subnet sizes to 'avoid waste'. In an IPv6 environment the only concern is the total number of subnets needed to meet routing/access policy, avoiding the nonsense of continually shifting the subnet size to align with the number of endpoints over time.

You also have to keep in mind that even if we could come up with a theoretically "perfect" address allocation scheme at minimum the existing space is going to be carved up 5 ways for each of the RIRs to implement. (When I was at IANA I actually proposed dividing it along the 8 /6 boundaries, which is sort of what has happened subsequently if you notice the allocations at 2400::/12 to APNIC, 2800::/12 to LACNIC and 2c00::/12 to AfriNIC.)

Since it's not germane to NANOG I will avoid rehashing the "why RA and 64-bit host IDs were bad ideas from the start" argument. :)

People need to get over it... the original design was 64 bits for both hosts and routing exceeding the design goal by >10^3, then routing wanted more so it was given the whole 64 bits. The fact that 64 more bits was added is not routing's concern, but the IPv4-conservation mindset can't seem to let it go despite having >10^6 more space to work with than the target. It could have been 32 bits (resulting in a 96 bit address), but given that 64 bit processors were expected to be widespread, it makes no sense to use less than that.

In the following I'm assuming that you're familiar with the fact that staying on the 4-byte boundaries makes sense because it makes reverse DNS delegation easier. It also makes the math easier.

I assume you meant 4-bit. ;) ^^^

As a practical matter we're "stuck" with /64 as the smallest possible network we can reliably assign. A /60 contains 16 /64s, which personally I think is more than enough for a residential customer, even taking a "long view" into consideration.

Stop looking backward. To achieve the home network of the last millennium a small number of subnets was appropriate. Constraining the world to that through allocation is a self-fulfilling way to force people that need more than that to develop/deploy hacks to work around the braindead practice.

The last time I looked into this there were several ISPs in Japan who were assigning /60s to their residential users with good success. OTOH, a /56 contains 256 /64s, which is way WAY more than enough for a residential user. The idea that a residential user needs a full /48 (65,536 /64s) is absurd.

A consumer is not a professional network operator and can't do 'well managed' subnet structures the way that this community would. To allow a plug-n-play consumer network, a fair amount of 'waste' is required.

OTOH, assigning a /48 to even a fairly large commercial customer is perfectly reasonable. This would give them 256 /56 networks (which would in turn have 256 /64 networks) which should be plenty to handle the problems of multiple campuses with multiple subnets, etc.

So let's assume that we'll take /56 as the standard unit of assignment to residential customers, and /48 as the standard unit of assignment to commercial customers. A /32 has 65,536 /48s in it. If your business was focused mainly on commercial customers that's not a very big pool.

A /32 is a starter-kit for a new ISP with no customers, so this argument is a continuation of the bogus 'denial' mindset used to avoid deployments. A real ISP should be documenting the number of customers they have * /48, then add in normal hierarchy loss and get a 'REAL BLOCK'.

OTOH if your business was focused primarily on residential customers you'd have 16,777,216 /56s to work with. That's enough for even a very large regional ISP. One could also easily imagine a model where out of a /32 you carve out one /34 for /56 assignments (4,194,304) and use the other 3/4ths of the space for /48s (49,152).

A really large ("national" or even "global") ISP would obviously need more space if they were going to intelligently divide up addresses on a regional basis. A /28 would have 16 /32s which should be enough for even a "very large" ISP, but let's really make sure that we cover the bases and go /24 (256 /32s). Even if you assume splitting that address space in half, that's 2.147483e+09 (approximately 2,147,483,000) /56s, and 8,388,608 /48s.

There are roughly 2,097,152 /24s in 2000::/3 (I say "roughly" because I'm ignoring space that's already been carved out, like 6to4, etc.), or 262,144 /24s per /6, or 67,108,864 /32s per /6. Which means that yes, we really do have "a lot" of space to work with. I also think it means that even with "a lot" of space there is no point in wasting it with foolish allocation policies that give out way more space than is realistically necessary just "because we can."

I agree we should not be foolish, but that includes not partitioning the space into a bunch of discontiguous blocks just 'because that is how we did IPv4'. From a big picture perspective we have already avoided 'waste' by agreeing to limitations on how much a new ISP gets and using a justification process for those that need more. The fact that end customers are not required to do the same level of justification (until they get really large), is avoiding waste in terms of process. The lack of tight control at the edge allows for innovation, which in turn drives the whole system forward.

I've ignored PI space up till now but I think it's reasonable for there to be a midpoint for PI somewhere between /48 and /32. Personally I think that a /40 has a nice sound to it. That's 256 /48 networks. I don't see any reason why the RIRs couldn't also agree to a /36, which would be 4,096 /48s. Even I don't see any reason why they should mess around with numbers like /41 or /43.

I personally agree, but we have a structure in place, and it has to have something to do to justify its continued existence.

To get back to the question that started the original thread, if I were the one who was requesting an IPv6 allocation I would use the following formula:

1 /56 per # of residential customers expected in 10 years + 1 /48 per # of business customers expected in 10 years

I would make that /48 & /44

Then assuming your current numbers are roughly 1/16th of what you hope they'll be in 10 years; when actually handing out addresses I'd give out the first /60 from each /56 to the residential customers. That way if you need to you can go back and chop up those /56s.

BS ... go back and get the larger block that supports the current customer base. The argument you present here says that an ISP gets one & only one shot at an allocation. The only way the approach of growing the request to match the customer base will ever be a problem is if every person on the planet individually subscribed to every service provider, but even that nonsense would work up to 32k providers globally.

I'd also start off handing out the first /48 out of every /44 to my commercial customers. That way they will have room to expand painlessly. This is sort of a bastardized version of the "sparse allocation" model that the RIRs have promoted. (Obviously the 1/16th number was chosen for convenience, but hopefully you get the idea of what I'm going for here.)

I realize that this is quite long, so if you've gotten this far, congratulations! I hope it was useful.

As people emerge from denial they need this type of lengthy and well reasoned line of thought, even though I disagree with some of the details. Tony