This archive is retained to ensure existing URLs remain functional. It will not contain any emails sent to this mailing list after July 1, 2024. For all messages, including those sent before and after this date, please visit the new location of the archive at https://mailman.ripe.net/archives/list/[email protected]/
[address-policy-wg] IPv6 addresses to transit-providers
- Previous message (by thread): [address-policy-wg] IPv6 addresses to transit-providers
- Next message (by thread): [address-policy-wg] IPv6 addresses to transit-providers
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Tim Streater
tim.streater at dante.org.uk
Wed Mar 2 14:30:42 CET 2005
At 10:32 02/03/2005, Jeroen Massar wrote: >On Wed, 2005-03-02 at 11:17 +0100, Hans Petter Holen wrote: >>Jeroen Massar wrote: >> >>>On Tue, 2005-03-01 at 17:31 +0000, Tim Streater wrote: >>> >>> >>>> 1) be an LIR - OK fine, we're an LIR. >>>> >>>>2) not be an End Site - OK we're not. >>>> >>>>3) plan to provide IPv6 connectivity to organisations - yes, we will certainly do that - to which it will assign /48s etc etc - no, we will never do that as all our customers are LIRs. >>>> >>>>4) have a plan for making at least 200 /48 assignments to other organisations etc etc - no, we will never assign such space as all our customers have their own already. >>>> >>>> >>> >>>According to the above, either 4 is false or 2 is false and you are >>>simply an endsite. Might sound harsh, but that is it... at the moment... >>> >>> >>Whait if I am (mainly) anIPv6 transit provider with 201 customers - all >>beeing LIR on their own: >>- I cannot get address space from my upstream because I have none or >>several depending on my size and definition of "up" >>- I cant make a plan to assigh 200 /48s since all my customers are LIRs >>on their own > >Do these customers are LIR's because they have 200 customers themselves or because being LIR allows them to get some address space more easily? >When they are endsites, that is having no other transits, they should be getting space from you and not themselves. In our case the customers are the NRENs of Europe so they certainly have >200 customers each. >>- I am hardly an end site ? >>how do I get adresses under the current policy ? > >This is I assume indeed the scheme that Tim from Dante shows. >And indeed this does not work under the current policy and as such that needs to be fixed.... > >The question boils down to: >- do you require a entry in the routing table >or: >- do you need address space > >Giving a /32 to such a site would be quite some waste, as you will never use it. >A /40 could be appropriate. But do you really need the entry in the routing table? If in fact all we do is assign space to the transit infrastructure, a smaller space could be sufficient. But (a) we need it to be routeable, else how can we manage/monitor it from entities outside the network (which we do), and (b) how do we get the space? Not from one of our customers, and we have no upstream. I am told that no prefix longer than /35 is routeable. >>If I cannot, how do we modify the policy to alow me to get adresses ? > >Propose a new/addition/change to the policy that specifies this specific case, bring it forward and let people vote. >If for one see why, especially in business case, you want your own address space. >For that matter a micro-policy would sort of be good, but the thing is..... how much routing entries will we end up with, again as much as with IPv4? > >>This is an excellent point to show were the addressing policies puts >>limitations on the structure of the ISP industry unless we are careful. We have PoPs from Vienna to Lisbon on a network with its own AS number. We do transit for the customers only. Everything else they can do for themselves and the customers set us up to do just what we do. With the existing policies I am not able to get v6 space for such a network. And why should not *any* group of entities decide to assemble a similar network, if they feel like it. If the v6 designers wanted to keep the routing table small, perhaps they should have gone for geographical prefixes like the phone system. -- Tim
- Previous message (by thread): [address-policy-wg] IPv6 addresses to transit-providers
- Next message (by thread): [address-policy-wg] IPv6 addresses to transit-providers
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]