The IANA (Internet Assigned Number Authority) distributes IPv6 address to RIR’s (Regional Internet Registry’s) around the world. At the moment there are five RIR’s and each of them is responsible for allocating IPv6 address space to ISP’s (Internet Service Providers) and, in some cases, End-User organizations. Once a block of addresses is allocated to an ISP it becomes their responsibility to distribute the address space to their customer base.
Let’s assume that an ISP is allocated a /32 by ARIN. In the early days of IPv6 it was often said that everyone would be given a /48 by their provider. And when I say ‘everybody’, I mean everybody, including residential households. Each /32 allows for 65,536 /48’s and each /48 allows for 65,536 /64’s. Because of what appears to be an almost infinitely abundant address space it seemed to make sense to keep things simple (e.g. give everybody a /48) and to eliminate the likelihood that an individual or an organization would actually run out of addresses. Now I have always loved that a design objective for IPv6 was to create an address space that had enough addresses so we no longer had to worry about addresses. I like the “let’s put this thing to bed for good” philosophy. It’s tantamount to choosing quality over price; pay for something of quality and it will last a lifetime but buy something cheap and you’ll have to buy it over and over during your life, ultimately paying more than you would have for quality. But even in my most wild and extravagant imaginings I can’t conjure uses for, much less a need for, 65,536 subnets at my house (my very own /48). This is especially true considering the fact that each of those 65,536 /64’s support more than 18.4 quintillion hosts. And in all seriousness, even if I could make up a way to use that many networks can you name a consumer who would have the financial resources to buy all of the gear necessary to build it? A /48 for everybody who wants one is excessive but it also accomplishes the objective of putting the “I’m running out if IP addresses” complaint to bed forever. And there are also some technical arguments regarding routing table size and hardware speed/efficiency that suggest it is inefficient to make prefixes smaller than /48.
I consider myself to be pretty geeky so ideas like IP-enabled milk cartons are incredibly exciting to me. But even when I sit down and dream up crazy ways to network my home I find it difficult to come up a need for more than a few dozen subnets. Chances are that my IP address needs would be forever satisfied with shortages never being a concern even if I had to struggle along with a /56. A /56 gives me 256 subnets to putter about with at my house and I can’t, for the life of me, think of ways that my house would need a 257th network. But I’m starting to push it when I suggest that the same is also true for a /60. With a /60 I will have 16 subnets to work with. And that seems a little too tight a space to work in for me. With a /60 I can see highly-technical homes having subnet issues.
It has long been assumed (by me) that ISP’s would balk at the idea of giving /48’s to their client base. If a single /48 can be carved into 256 /56’s and few to no customers are going to complain about having to solve their networking needs with a /56 it only makes sense that ISP’s would do it. Everybody is technically satisfied and the ISP’s can hoard their other /48’s for future use. And by “future” I mean that they would probably never use them. The decision on whether or not your ISP was going to give you a /48, /56, /60 or even a /64 was going to be between you and the ISP; the RIR’s had nothing to do with it.
And then someone suggested that ARIN change section 18.104.22.168 of their allocation policy document from this:
22.214.171.124. Assignment address space size
End-users are assigned an end site assignment from their LIR or ISP. The exact size of the assignment is a local decision for the LIR or ISP to make, using a minimum value of a /64 (when only one subnet is anticipated for the end site) up to the normal maximum of /48, except in cases of extra large end sites where a larger assignment can be justified.
The following guidelines may be useful (but they are only guidelines):
/64 when it is known that one and only one subnet is needed
/56 for small sites, those expected to need only a few subnets over the next 5 years.
/48 for larger sites
RIRs are not concerned about which address size an LIR/ISP actually assigns. Accordingly, RIRs will not request the detailed information on IPv6 user networks as they did in IPv4, except for the cases described in Section 6.4.4 and for the purposes of measuring utilization as defined in this document.
LIR’s may assign blocks in the range of /48 to /64 to end sites.
All assignments made by LIR’s should meet a minimum HD-Ratio of .25.
* /64 – Site needing only a single subnet.
* /60 – Site with 2-3 subnets initially.
* /56 – Site with 4-7 subnets initially.
* /52 – Site with 8-15 subnets initially.
* /48 – Site with 16+ subnets initially.
LIR’s do not need to issue all 5 sizes of prefixes as long as the
HD-Ratio requirement is met.
Note: An explanation of HD ratio can be found in RFC 3194.
Many people took exception to this suggested wording and claim that is smacks of ARIN trying to tell ISP’s how to distribute their address space. Other people feel that this makes complete sense because it is a much more conservative approach. Most of the latter continue to suffer from the aftershocks of IPv4’s address issues and they can’t do anything other than apply their past thoughts to this new approach.
The reality is that there is so much address space available that every living soul on earth today will have long since died before we can even begin to think about putting pressure on the IPv6 address space. So why? Why? Why are worrying so much about conserving when the single biggest thing it’s going to do is make routing tables larger, subnetting more prone to error and routing hardware less efficient? There has to be an argument more compelling than, “It’s wasteful.”
As of today the wording has not been adopted and I hope it stays that way. ISP’s are fully capable of figuring this stuff out on their own.