[6bone] proposal for transfer of 6bone address management responsibilities to RIRs

Tim Chown tjc@ecs.soton.ac.uk
Sat, 24 Aug 2002 15:57:42 +0100


On Sat, Aug 24, 2002 at 09:23:30AM +0200, Gert Doering wrote:
> Hi,
> 
> The extra price comes from distinguishing "ultra-low commodity products"
> and "product that requires some customer-specific configuration" - the
> latter should be less effort for IPv6...

Agreed, the question is single static IPv4 IP cost vs static /48 assignment
cost.  Both have "some customer-specific configuration".   Also, I'm not 
sure the prefix delegation mechanisms have been fully agreed yet? (it was
being discussed in IETF in March, but I haven't tracked this specific topic
since then).    

If you compare /48's (the future :) with single IPv4 addresses allocated by
ISPs now, an ISP getting a /32 is going to have (if you ignore the HD ratio)
65K customer prefixes, which in essence lets them serve as many customers
as a provider with (in old money) a Class B v4 allocation.   It's not a
*lot* of space so while the provider can go back to get the /32 made bigger,
they may be more tempted to assign the /48's from a pool.   The difference
being with v6 it is much easier to run services into the customer network,
so the demand for static addresses (or a very scalable DDNS :) would appear
higher.   If everyone offers it, I guess there won't be a premium.

Tim