[WARNING: This article may include language that may offend Service Providers, and include terms that are incomprehensible to the non-IPv6 aware]
Today while I was discussing today about how Service Providers should allocate IPv6 addresses to home-user type customers, I found this RFC (dated March 2011)
Specifically, it modifies the IETF’s previous stance that ALL customer allocations be /48. Specifically:
“this document still recommends giving home sites significantly more than a single /64, but does not recommend that every home site be given a /48 either” (RFC6177)
It more-or-less recommends that home sites be given a /56:
“The above-mentioned goals of RFC 3177 can easily be met by giving home users a default assignment of less than /48, such as a /56.” (RFC6177)
Here’s hoping the service providers get behind the recommendation, and don’t start handing out /64 addresses to home or SOHO users.
So there you have it. The latest published “best current practice” for IPv6 allocation suggests /56 be given to home users when allocated IPv6 addresses.
My suspicion is that Service Providers will still be feeling scared as a hangover from IPv4 days, and NOT follow the best current practice, and will quite likely only allocate /64 prefixes to home users. Shame on you if you do!
Background (for the not-so-technically minded)
IPv6 addresses are 128 bits long, and in RFC3177 it was recommended that:
“Home network subscribers, connecting through on-demand or always-on connections should receive a /48”
Which means that if a service provider had been allocated all of the IPv6 addresses that begin with 2001:0DB8, (ie is a /32 block – the size which is recommended is given to service providers) which can be thought of as 2001:0DB8:xxxx:xxxx:hhhh:hhhh:hhhh:hhhh (the hhhh bits represent the interface address, and can’t be changed) they should allocate them in blocks of 48 bit prefixes, so the first block would be:
2001:0DB8:0000:xxxx:hhhh:hhhh:hhhh:hhhh; where the xxxx values could be used by the customer to give themselves up to 65,536 subnets. The second block would be
2001:0DB8:0001:xxxx:hhhh:hhhh:hhhh:hhhh; then
2001:0DB8:0002:xxxx:hhhh:hhhh:hhhh:hhhh; and so on to
2001:0DB8:FFFF:xxxx:hhhh:hhhh:hhhh:hhhh
This limits service providers to a mere 65,536 customers before they have to go back to their RIR and ask for a new allocation.
RFC6177 modifies this so that the Best Current Practice is to now give home users allocations of /56; so reworking the example above, the first block would be:
2001:0DB8:0000:00xx:hhhh:hhhh:hhhh:hhhh; where the xx values could be allocated by the customer, allowing them 256 subnets if required. The second block would be
2001:0DB8:0000:01xx:hhhh:hhhh:hhhh:hhhh; then
2001:0DB8:0000:02xx:hhhh:hhhh:hhhh:hhhh; and so on.
This means that Service Providers won’t need to go back to the RIR until they have services 16,777,216 such home customers.
BUT, I suspect that Service Providers will be greedy. And want to make it difficult for home users and SOHO users to use any subnets at all, so will in defiance to RFC6177, allocate /64 prefixes to customers, thereby denying them the ability to create any subnets at all.
Thanks , I’ve just been looking for info about this topic for ages and yours is the best I’ve discovered till now. But, what about the bottom line? Are you sure about the source?
I’m dead sure about the RFCs. And you might glean a bit more form this too: http://www.ripe.net/ripe/docs/ripe-552
Pingback: Just how many IPv6 addresses are there? Really? | RedNectar's Blog
Pingback: When will the IPv6 address pool be depleted? « Reggle
Nice article Reggle. I’m with you.
Thanks! Nice blog, you too!