AW: [ipv6-wg] Re: Re: [address-policy-wg] Re: Andre's guide to fi x IPv6
Hi Gerd,
-----Ursprungliche Nachricht----- Von: Gert Doering [mailto:gert@space.net] Gesendet: Montag, 5. Dezember 2005 18:07 An: Kurt Erik Lindqvist Cc: Cameron C. Gray; Per Heldal; address-policy-wg@ripe.net; ipv6-wg@ripe.net Betreff: Re: [ipv6-wg] Re: Re: [address-policy-wg] Re: Andre's guide to fix IPv6
Hi,
On Mon, Dec 05, 2005 at 02:17:25PM +0100, Kurt Erik Lindqvist wrote:
I think each LIR should get a /32 and we should drop the 200 "customer" rule. But that is just me...
Actually I like the "every AS should get a <network>" approach (with a yearly recurring fee for AS and network, to guarantee return of the resources as soon as the importance of having a slot in the global routing table doesn't outweigh the costs anymore).
"Every LIR gets a /32 (upon request), no questions asked" is a concept that I'm also happy with - as I have said before. For those that are afraid of the landrush: limit that policy to 5.000 LIRs per region.
Not sure if that works. Kind of "First come - first serve" thing. Regards Olaf
Gert Doering -- NetMaster -- Total number of prefixes smaller than registry allocations: 81421
SpaceNet AG Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 D- 80807 Muenchen Fax : +49-89-32356-234
Hi, On Mon, Dec 05, 2005 at 06:13:03PM +0100, Bonness, Olaf wrote:
"Every LIR gets a /32 (upon request), no questions asked" is a concept that I'm also happy with - as I have said before. For those that are afraid of the landrush: limit that policy to 5.000 LIRs per region.
Not sure if that works. Kind of "First come - first serve" thing.
Yes, so what shall we do instead? "Nobody gets addresses, no mistakes made"? The whole point of this discussion is that there are two fractions: - one side is so afraid of doing the wrong thing that they want to delay everything until the perfect solution is found - the other side wants to get things rolling, and is willing to risk a few thousand entries in the global BGP table (whatever that is). and I don't think we'll ever reach consensus between those groups. Gert Doering -- NetMaster -- Total number of prefixes smaller than registry allocations: 81421 SpaceNet AG Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 D- 80807 Muenchen Fax : +49-89-32356-234
Gert Doering wrote:
Hi,
On Mon, Dec 05, 2005 at 06:13:03PM +0100, Bonness, Olaf wrote:
"Every LIR gets a /32 (upon request), no questions asked" is a concept that I'm also happy with - as I have said before. For those that are afraid of the landrush: limit that policy to 5.000 LIRs per region. Not sure if that works. Kind of "First come - first serve" thing.
Yes, so what shall we do instead?
"Nobody gets addresses, no mistakes made"?
2000::/3 is only 1/8th of the total IPv6 address space, we can make mistakes 8 times :) But the problem then is that we will have 8 swamps of crappy managed address space which most likely nobody will return. For that matter, there is not much force to get folks to return address space but having the parties they want to talk with filter it out. I wonder what would happen if eg Google would ask (access) ISP's to start paying for traffic 'or else' :)
The whole point of this discussion is that there are two fractions:
- one side is so afraid of doing the wrong thing that they want to delay everything until the perfect solution is found
- the other side wants to get things rolling, and is willing to risk a few thousand entries in the global BGP table (whatever that is).
and I don't think we'll ever reach consensus between those groups.
Especially as there are no proper argumented proposals being made by the second group, except that the current one is not to their liking and that things that get proposed doesn't fit their bill. My view on all of this is the following (hinting group #2 :) : - reserve a global, thus all the RIR's in one slot and split that up under the RIR's, block for 'multihoming purposes' (3ffe::/16 seems very appropriate as it is already swamp) - give people who want to multihome a /48(*) out of this block. - let them pay an annual fee for it that makes it a real incentive to maintain it and use it. - let them sign a document that when proper multihoming (eg SHIM6) comes that they will start using that and after X time stop announcing their /48 more specific. Also let other people filter at wish. (Add lawyer style wording in the above ;) * = or other appropriate size, maybe per /40 is better so that each multihomer could have 256 sites behind them, though then they are at the 200-rule which they seem to be bothered with already... :) This gives them PI address space, when SHIM6 or a similar method gets realized the swamp gets cleaned. Of course the swamp will only be cleansed when everybody cooperates and I don't see that happening. One advantage of the "multihoming block" is that it will be very easy to filter out. Of course such multihoming blocks can have private peerings and people that don't filter them won't see them. But HEY isn't that exactly what is already happening now? Check GRH* to see how many /48's from separate ASN's can already be seen (~60 at the moment). I guess that 6bone space will be a perfect example on what happens with swamp space that is supposed to be returned. The big issue in the above is that many people will not like SHIM6 or other methods because it is not what they have now and they will claim that they don't have proper control over the route announcements of that block. I am then inclined to ask them how much control they have over their route announcements when it leaves their ASN... ;) Greets, Jeroen * = http://www.sixxs.net/tools/grh/lg/?show=endsite&find=::/0
Hi, On Tue, Dec 06, 2005 at 12:32:44AM +0100, Jeroen Massar wrote:
- the other side wants to get things rolling, and is willing to risk a few thousand entries in the global BGP table (whatever that is).
and I don't think we'll ever reach consensus between those groups.
Especially as there are no proper argumented proposals being made by the second group, except that the current one is not to their liking and that things that get proposed doesn't fit their bill.
That's not fully correct. There are a number of fairly specific proposals (to name three, in the order of "radicality": "abandon 200-customer rule", "give every LIR a /32", "give every AS holder a prefix"). Gert Doering -- NetMaster -- Total number of prefixes smaller than registry allocations: 81421 SpaceNet AG Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 D- 80807 Muenchen Fax : +49-89-32356-234
Bonness, Olaf wrote:
Von: Gert Doering [mailto:gert@space.net]
"Every LIR gets a /32 (upon request), no questions asked" is a concept that I'm also happy with - as I have said before. For those that are afraid of the landrush: limit that policy to 5.000 LIRs per region.
Not sure if that works. Kind of "First come - first serve" thing. Regards
Also, such a policy is most likely in direct conflict with one of the basic paradigms of a RIR - which is equal treatment of its members in equal circumstances. Best, -C.
Hi, On Mon, Dec 05, 2005 at 07:14:33PM +0100, Carsten Schiefner wrote:
Von: Gert Doering [mailto:gert@space.net]
"Every LIR gets a /32 (upon request), no questions asked" is a concept that I'm also happy with - as I have said before. For those that are afraid of the landrush: limit that policy to 5.000 LIRs per region.
Also, such a policy is most likely in direct conflict with one of the basic paradigms of a RIR - which is equal treatment of its members in equal circumstances.
Circumstances before and after reaching a given number of routes are not "equal". Right now, nobody knows whether we will *ever* reach 10.000 IPv6 routes in the global BGP table - if we hit 10.000, we know that there is a given possibility of having "very very very many routes", and that conservation of routing table slots might be more of an issue than right now (where the issue is more along the lines of "get it going, or abandon it"). Gert Doering -- NetMaster -- Total number of prefixes smaller than registry allocations: 81421 SpaceNet AG Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 D- 80807 Muenchen Fax : +49-89-32356-234
Hi Gert, Gert Doering wrote:
On Mon, Dec 05, 2005 at 07:14:33PM +0100, Carsten Schiefner wrote:
Also, such a policy is most likely in direct conflict with one of the basic paradigms of a RIR - which is equal treatment of its members in equal circumstances.
Circumstances before and after reaching a given number of routes are not "equal".
where _eaxctly!_ sits the difference between LIR #5000 and LIR #5001 - to pick up your example? And why is '5,000' _not!_ totally arbitrary? Although I am mostly agnostic wrt. the general discussion here so far, I am absolutely in line with Andre Oppermann's comment that we better have profound and sound answers to those questions ready before such a policy would be put in place - because they would be asked the second after anyways, most likely by these well-known, interested third parties... Best, Carsten
Hi, On Tue, Dec 06, 2005 at 12:12:32PM +0100, Carsten Schiefner wrote:
Gert Doering wrote:
On Mon, Dec 05, 2005 at 07:14:33PM +0100, Carsten Schiefner wrote:
Also, such a policy is most likely in direct conflict with one of the basic paradigms of a RIR - which is equal treatment of its members in equal circumstances.
Circumstances before and after reaching a given number of routes are not "equal".
where _eaxctly!_ sits the difference between LIR #5000 and LIR #5001 - to pick up your example?
It's like growing up, and being 18 years old, over night. Nothing but a small number change, but lots of effect.
And why is '5,000' _not!_ totally arbitrary?
I didn't say that it's not completely arbitrary. It's a limit that is loosely in the same range as the current number of LIRs per RIR, and well below the limits that will cause problems for currently deployed router architectures. I don't seriously see us reaching this number anyway in the near future - but I've put it there in case that I'm wrong, and those people that are afraid of a massive land rush are right.
Although I am mostly agnostic wrt. the general discussion here so far, I am absolutely in line with Andre Oppermann's comment that we better have profound and sound answers to those questions ready before such a policy would be put in place - because they would be asked the second after anyways, most likely by these well-known, interested third parties...
I'm open for any proposal that will fulfill all the necessary criteria (in no specific order, and I'm sure I forgot one or two): - fair - not-ITU-problematic - not-monopoly-problematic - not using up BGP routing table slots - giving anybody maximum independence from anything else in the Internet - not wasting address space - giving anybody maximum address space so that they can be maximum convenient (internal hierarchy and allocation, etc.) - having a one-size-fits-all model for end users, to discourage bureaucracy - giving machines on a link the benefit of autoconfiguration - not wasting a /64 on links, and /48s on users as you can easily see, there is no way any policy can fulfill all of this at the same time. So the only thing that can ever reach consensus is a foul compromise (or a miracle from the IETF). "Foul compromise" = "nobody is happy with it, but everybody can *live* with it". Gert Doering -- NetMaster -- Total number of prefixes smaller than registry allocations: 81421 SpaceNet AG Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 D- 80807 Muenchen Fax : +49-89-32356-234
participants (4)
-
Bonness, Olaf -
Carsten Schiefner -
Gert Doering -
Jeroen Massar