[AusNOG] IPv6 Addressing

[John.Gibbins at csiro.au]

We all agree that the numbers involved with IPv6 are huge.  It is conceivable that with wasteful allocation policies it would be possible to run out.  This can be compared with the old A-Class allocations.  There were good reasons at the time for allocating blocks based on classes but this was phased out as technology improved.

I think a key point with IPv6 is that most of the addresses are reserved.  All current routable unicast addresses are in the 2000::/3 block (1/8th of the address space).  Multicast and local addresses are in the f000::/4 block, even with a few other minor reservations that still leaves over 3/4 of the total address space free to be allocated more efficiently in future if we ever get to that stage.

Internally we do something similar with our allocation.  We reserved the top 4 bits of our /32 so we can implement a different allocation policy in future (probably never, but who can predict).  In the mean time we happily allocate /64s to all networks including point to point links and even single loopback addresses on routers.

Regards
johng
PS My email address is accessible via IPv6!
--
John Gibbins
IT Security Operations (and IPv6 Evangelist)
CSIRO Information Management & Technology (IM&T)
Phone: +61 2 6124 1419  | Fax: +61 2 6124 1414 | Mob: 0419 605 562
John.Gibbins at csiro.au | www.csiro.au 
PO BOX 225, Dickson  ACT 2602
Do not let what you cannot do interfere with what you can do - John Wooden

-----Original Message-----
From: ausnog-bounces at lists.ausnog.net [mailto:ausnog-bounces at lists.ausnog.net] On Behalf Of Jones, Rick
Sent: Thursday, 7 April 2011 8:05 AM
To: ausnog at lists.ausnog.net
Subject: Re: [AusNOG] IPv6 Addressing

I think we can safely say that the magnitude of the numbers are such that the problem will not surface for a very long time, and the technology to manage it will be very, very different.  I know it sounds like we are putting our heads in the sand, but in fact we are not.

Ipv4 is able to deliver 1,677,216 /24 subnets (ignoring reserved addresses).
Ipv6 is able to deliver 18,446,744,073,709,600,000 /64 subnets (again ignoring reserved addresses).

Making the subnet smaller gives us very little practical benefit - CPUs like working with numbers on 32 bit boundaries and going to 79,228,162,514,264,300,000,000,000,000 /32 subnets gives us no practical benefit at all, unless we want to assign addresses at the atomic or sub-atomic level.

The benefit of /64 is that it is consistent across the Internet, and means that we can forget about the sizes of subnets once and for all.  The order of magnitude jump from 32 bits to 128 bits gives us 79,228,162,514,264,300,000,000,000,000 ipv4 address spaces to play with, far more than we are ever going to use in the next 40 years.

Cheers,
Rick

*********************************************************************