Leo Vegoda wrote:
Hi Jordi,
On Jan 26, 2007, at 10:04 AM, JORDI PALET MARTINEZ wrote:
Hi Leo,
Yes, good point, that's the idea.
Are you trying to limit the commercial freedom of ISPs to offer a range of products based around different sized networks?
Well, I think it is inevitabale to revisit that issue and to reopen that case. For 2 reasons: - address space was never to be "sold" to end users by "volume", it was always to be provided on an as needed basis. Address space is not an ISPs "property" to be resold for profit. There are many other (better) ways to implement commercial freedom and product differentiation. In the IPv4 world, the (marketing departments of) ISPs got away with (mis)using the scarcity argument for wiring the size of address blocks into the packages they are offering. My personal view on that: the users got used to it and the community, including the RIRs were/are tolerating this approach. - for IPv6, one of the explicit goals (architectural assumptions) was that the community should do away with the bickering, the mechanisms that support or allow IPv4-like ISP lock-in mechanisms, by *assigning* equally sized blocks (/48) to end sites because IPv6 addresses are plenty. Other than trying to limit the attractiveness of "selling" addresses, there are more reasons why a homogenous address distribution seems favourable: When moving site networks around (ISP a -> ISP b) the site would not have to re-design the subnet layout, just the LHS bits in the addresses would have to be changed. (Anyone still remembering A6 RRs to make that easy? ;-) ) Now if/when we encourage the situation that different ISPs offer different "packages", it is easy to end up in a situation where a site would be holding a say /48, but another ISP would (by default) only assign /56. Then, either the site would have to re-structure its subnet landscape or *and that's the point here* should have the right to request *and* to receive an equally sized and *contiguous* assignment from ISP b, up to the originally proposed /48. Anythig else is (again, and through the same backdoor of "conservation") just a motivation again to "sell" address space again. And to promote address translation mechanisms and products...
I remember this issue being discussed in the context of IPv4 when ripe-152, the 'Charging by Local Internet Registries' document was reviewed a couple of years ago. I seem to remember several strong statements against RIR policy regulating business models during that discussion. Is this a discussion you want to re-visit?
In the context of this proposal, I think we need to do that, otherwise we are pulling the carpet away from underneath the original architectural assumptions and recommendations for IPv6.
http://www.ripe.net/ripe/wg/address-policy/r47-minutes.html (Section F)
Regards,
The problem(?) that I'm seeing here is that the RIPE community is made up of more ISPs than end-users and customers. So there may well be a bias to preserve the IPv4 business models for the IPv6 world. So, when you are asking "Are you trying to limit the commercial freedom of ISPs", from a communities point of view and my personal one: yes, *in this particular respect* because there are plenty of other and more appropriate mechanisms to use for product differentiation. Wilfried