Kurt Erik Lindqvist [mailto:kurtis@kurtis.pp.se] wrote:

On måndag, maj 12, 2003, at 21:33 Europe/Stockholm, Gert Doering wrote:

...

...

500 TLA's over 54 countries!

Oh, wow. Very impressive, this!

Maybe we actually will hit 1000 rotues! :-)

Another thing though :

4 790 829408 440969 18406 0 0 2w5d 424

I only see 424 of these though. Anyone that have already started mapping the times of when an allocation is made and when it first shows up in the routing table?

http://www.sixxs.net/tools/grh/tla/ With last seens atm, got to fish the first seens from the archive (show up as empty now). They will pop up later today :) Note that a 'first seen' is not very useful as it could pop up for a couple of minutes (even due to routing troubles) and then fade away again... When you follow the link behind the 'lastseen' you will see all the marks made where the prefix was originating from and over how many ASPaths it was detected. Greets, Jeroen

% >http://www.sixxs.net/tools/grh/tla/ % > % >With last seens atm, got to fish the first seens from % >the archive (show up as empty now). They will pop up % >later today :) Note that a 'first seen' is not very useful % >as it could pop up for a couple of minutes (even due to % >routing troubles) and then fade away again... % >When you follow the link behind the 'lastseen' you will see % >all the marks made where the prefix was originating from % >and over how many ASPaths it was detected. % % This keeps turning into an ever more useful page. Kudos to the author! % % What are the RIRs doing to reclaim all those "red" lines of allocations % that have never once appeared in a routing table? I would say all % assignments from before Jan 2002 and that have "never" in the "last seen" % column are ripe for revocation. Question is, why hasn't this been done % before and has this been discussed so far? % % -Hank Well, in at least one case, 2001:0478, you should never see a /32 or /35 announcement. This prefix is used for exchange points and critical infrastructure and is delegated as /48s --bill Opinions expressed may not even be mine by the time you read them, and certainly don't reflect those of any other entity (legal or otherwise).

On Fri, 16 May 2003, leo vegoda wrote:

Hank Nussbacher <hank@att.net.il> writes:

[...]

...

What are the RIRs doing to reclaim all those "red" lines of allocations that have never once appeared in a routing table?I would say all assignments from before Jan 2002 and that have "never"in the "last seen" column are ripe for revocation.Question is, why hasn't this been done before and has this been discussed so far?

Is there a policy requirement that IPv6 prefixes allocated by RIRs must be routed on The Internet? I've not found it in the current "IPv6 Address Allocation and Assignment Policy".

Is there such a requirement for IPv4 prefixes? If yes (and I would hope so, otherwise why would anyone want RFC1918 addresses when one can get "real" IPs), then I think the same should apply for IPv6 prefixes.

Regards,

-- leo vegoda RIPE NCC Registration Services

-Hank

On Fri, 16 May 2003, Bill Manning wrote:

% > Is there a policy requirement that IPv6 prefixes allocated by RIRs must % > be routed on The Internet? I've not found it in the current "IPv6 % > Address Allocation and Assignment Policy". % % Is there such a requirement for IPv4 prefixes?If yes (and I would hope % so, otherwise why would anyone want RFC1918 addresses when one can get % "real" IPs), then I think the same should apply for IPv6 prefixes. % % > leo vegoda % % -Hank

there has never been a requirement that a prefix berouted or announced on the "Internet".

In that case why does an ISP have to: 5.1.1.d: have a plan for making at least 200 /48 assignments to other organisations within two years. See: http://www.ripe.net/ripe/docs/ipv6policy.html -Hank

--bill Opinions expressed may not even be mine by the time you read them, and certainly don't reflect those of any other entity (legal or otherwise).

Hank Nussbacher

Hi, On Fri, May 16, 2003 at 05:31:45AM -0700, Bill Manning wrote:

there has never been a requirement that a prefix be routed or announced on the "Internet".

Yes, and 32 bits are enough for everybody :-) So maybe now is the time to look upon the way things have been done in the past and consider "is that the way we want to do them in the future"? The IPv6 policy as it stands now (which doesn't say that it's cast in stone or that it's a perfect policy - beware) gives IPv6 address space to entities that claim that they are going to use it to facilitate internet access for (200 and more) 3rd parties. So if that address space isn't visible, the prerequisites are not fulfilled, obviously, and it would be in the boundaries of the policy to take the address space back. As of today, I do not think that's useful. Why? Because people might just be slow in building their IPv6 networks, or have put their projects on hold (due to financial reasons). So being overly restrictive here is just hurting IPv6 deployment, for no gain. Something worth to do for someone with too much time on their hand is to figure out whether those companies that have non-visible address space actually still exist, or whether they went under - in which case it would kind of "automatically" fall back to the registry. But then this is certainly not a high priority job for the *registries* - they have more important forward thinking to do. Gert Doering -- NetMaster -- Total number of prefixes smaller than registry allocations: 54495 (54267) SpaceNet AG Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 80807 Muenchen Fax : +49-89-32356-299

Hi, On Fri, May 16, 2003 at 06:20:24AM -0700, Bill Manning wrote:

% So if that address space isn't visible, the prerequisites are not % fulfilled, obviously, and it would be in the boundaries of the policy % to take the address space back. Visable to whom?

To the majority of the internet users (as you insist on claiming that there is nothing as "the global routing table"). The Internet is an *Inter*network. It's about connecting all of it together, not building small splinter networks that have no connectivity. There is currently no provision in the IPv6 policy for people that just want some local/VPN connectivity and no global routing. Maybe that needs changing as well. (Site-locals had some potential for "local" things, but they are dead, as far as I understand). David, could you put that on your list of things to consider? Gert Doering -- NetMaster -- Total number of prefixes smaller than registry allocations: 54495 (54267) SpaceNet AG Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 80807 Muenchen Fax : +49-89-32356-299

Hi, On Fri, May 16, 2003 at 10:17:11AM -0400, Kimmo Suominen wrote:

I've used IPv4 space in the past to connect to extranet providers using unique addresses, without connecting to the Internet. Is such use not allowed in the IPv6 world? One must promise to advertise the addresses to the Internet to get an allocation? End to extranet providers?

I wouldn't go so far as to say "it's not allowed". What I am saying is that the current IPv6 allocation policy was made with the needs of people in mind that want to connect to "the global Internet", so there is no clear answer how to fulfill those people's needs. When the policy was made, people were still suggesting the use of site-local addresses for "non-global" usage. Site-locals seem to be dead, so there is a hole in the policies right now. Passing on the question from the registry point of view to the IETF people (Michael & co): what are your recommendations how this can be addressed (in the double sense)? Gert Doering -- NetMaster -- Total number of prefixes smaller than registry allocations: 54495 (54267) SpaceNet AG Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 80807 Muenchen Fax : +49-89-32356-299

Hi, On Fri, May 16, 2003 at 03:39:53PM +0100, Carlos Morgado wrote:

On Fri, May 16, 2003 at 04:22:07PM +0200, Gert Doering wrote:

...

What I am saying is that the current IPv6 allocation policy was made with the needs of people in mind that want to connect to "the global Internet",

Not even that, as it doesn't take into account transit providers.

Yes, I'm aware of that. Most of those do have some sort of end customers, though, and should easily be able to get address space under the current policy. For those that can't, the policy needs to be adopted. The to-be-formed editorial committee needs to find a solution here. Gert Doering -- NetMaster -- Total number of prefixes smaller than registry allocations: 54495 (54267) SpaceNet AG Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 80807 Muenchen Fax : +49-89-32356-299

Hi, On Fri, May 16, 2003 at 08:11:31PM +0100, Carlos Morgado wrote: [..]

In my particular case it would be something of a stretch. We are almost a pure transit provider, most of our clients have their own address space and are multihomed. The ones that are using delegations from our space are fairly small.

Size of customer doesn't matter (everybody who might have two or more network segments gets a /48).

Also, it's a chicken/egg problem. I can draw up a plan with planned allocations for current customers but I can't commit to customer deploy dates - I can't even commit to they being our costumers a month from now.

This is well understood by RIPE hostmasters. Right now, the rules are meant to be interpreted in a very relaxed way - this is: if you have 200 or more IPv4 customers (that have addresses from you) that might go to IPv6 *if* it really takes off in the next years, then you qualify. If you have significantly below 200 IPv4 customers, it's becoming difficult - but you should nevertheless talk to the RIPE NCC hostmasters and see what could be done. (If it really doesn't work out under the current policy, you could use a /48 from one of your customers to number your infrastructure and your services - yes, this sucks, but it's better than no IPv6 transit at all) [..]

...

For those that can't, the policy needs to be adopted. The to-be-formed editorial committee needs to find a solution here. I'm sorry I missed the ipv6-wg meeting, was this problem discussed ?

It's more a lir-wg problem (as it's policy, not technical aspects). This specific problem wasn't discussed in either working group, but we are aware of it. There are a couple of other cases as well.

I'm also curious to know about other providers sufering from this problem and what solutions they propose.

Indeed. Comments welcome. Gert Doering -- NetMaster -- Total number of prefixes smaller than registry allocations: 54495 (54267) SpaceNet AG Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 80807 Muenchen Fax : +49-89-32356-299

On Sat, May 17, 2003 at 08:28:02PM +0200, Gert Doering wrote:

Hi,

On Fri, May 16, 2003 at 08:11:31PM +0100, Carlos Morgado wrote: [..]

...

In my particular case it would be something of a stretch. We are almost a pure transit provider, most of our clients have their own address space and are multihomed. The ones that are using delegations from our space are fairly small.

Size of customer doesn't matter (everybody who might have two or more network segments gets a /48).

Yes, that is indeed one scenario. But consider the extreme case of an operator which only sells transits to multihomed customers who already have their own address. Unless that operator starts assigning /48s to P-t-P costumer links he'll be hard depressed to come up with 200 nets qualifying as end site on his own infrastructure. This kind of network is not very diferent from a very big IX, except it needs to be globaly routable (you know what I mean, don't pick at there is no global routing table ;))

...

Also, it's a chicken/egg problem. I can draw up a plan with planned allocations for current customers but I can't commit to customer deploy dates - I can't even commit to they being our costumers a month from now.

This is well understood by RIPE hostmasters. Right now, the rules are meant to be interpreted in a very relaxed way - this is: if you have 200 or more IPv4 customers (that have addresses from you) that might go to IPv6 *if* it really takes off in the next years, then you qualify.

Or, if the smaller clients with PA IPv4 space have totally diferent time frames than the bigger PI transit clients. Anyway, I see what you mean.

If you have significantly below 200 IPv4 customers, it's becoming difficult - but you should nevertheless talk to the RIPE NCC hostmasters and see what could be done.

(If it really doesn't work out under the current policy, you could use a /48 from one of your customers to number your infrastructure and your services - yes, this sucks, but it's better than no IPv6 transit at all)

While being possible technically, I fear this is highly impossible :)

[..]

...

...

For those that can't, the policy needs to be adopted. The to-be-formed editorial committee needs to find a solution here. I'm sorry I missed the ipv6-wg meeting, was this problem discussed ?

It's more a lir-wg problem (as it's policy, not technical aspects).

I figure interested people are in both lists so I tried to keep cross posting down ;) -- Carlos Morgado <chbm@cprm.net> - Internet Engineering - Phone +351 214146594 GPG key: 0x75E451E2 FP: B98B 222B F276 18C0 266B 599D 93A1 A3FB 75E4 51E2 The views expressed above do not bind my employer.

Hi, to followup on myself (sorry for that): On Sat, May 17, 2003 at 08:28:02PM +0200, Gert Doering wrote:

If you have significantly below 200 IPv4 customers, it's becoming difficult - but you should nevertheless talk to the RIPE NCC hostmasters and see what could be done.

This is still my basic recommendation: talk to the NCC hostmasters, and *iff* they reject your application, come and talk to us (lir-wg) so that we can fix the policy, if necessary (but this is a slow process). Concerning the original question, a new idea just popped up that would solve the problem of the "big transit providers" - we could adapt the "200 customers" clause to "200 /48 assignments to customers, OR provide transit to 20 other ASes". Of course we might end up in dropping the "200 /48s" rule altogether, but this is pretty much open at this stage. If you want to participate in the IPv6 policy making process - and all of you are very welcome, of course! - please join the "global-ipv6" list (global-v6@lists.apnic.net, send mail to Majordomo@lists.apnic.net with "subscribe global-v6" in it). Gert Doering -- NetMaster -- Total number of prefixes smaller than registry allocations: 54495 (54267) SpaceNet AG Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 80807 Muenchen Fax : +49-89-32356-299

Hi, On Fri, May 16, 2003 at 08:09:46AM -0700, Bill Manning wrote:

% On Fri, May 16, 2003 at 06:20:24AM -0700, Bill Manning wrote: % > % So if that address space isn't visible, the prerequisites are not % > % fulfilled, obviously, and it would be in the boundaries of the policy % > % to take the address space back. % > Visable to whom? % % To the majority of the internet users (as you insist on claiming that % there is nothing as "the global routing table").

does that majority have to include you or I?

Of course not - if I disconnect my PC, I can't reach anyone. But this side discussion isn't helpful in any way.

and show me the global routing table please?

This is kinda difficult, as everybody has a local view of it, of course.

% The Internet is an *Inter*network. It's about connecting all of it % together, not building small splinter networks that have no connectivity.

The Internet in one mesh of interconected networks that run the IP protocol suite. There are others. Military networks, Closed commercial networks, Financial networks, Research networks, ... its a long list.

So what? If those networks decide to use different rules for IP/IPv6 address allocation and usage, why should we care? If they decide to become part of "The Internet", then they are part of the global routing table/system. Sorry, but I don't get your point. Gert Doering -- NetMaster -- Total number of prefixes smaller than registry allocations: 54495 (54267) SpaceNet AG Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 80807 Muenchen Fax : +49-89-32356-299

On Thu, 22 May 2003, Bill Manning wrote:

% So what? % % If those networks decide to use different rules for IP/IPv6 address % allocation and usage, why should we care? % % If they decide to become part of "The Internet", then they are part of % the global routing table/system.

Hum... where to begin. First off, it seems that you are making the assertion that entities will make the unconnected/connected transition -once- which emperical evidence suggests is not always true. In the past decade, there is a significant body of evidence that networks and nodes are gaining mobility. part of that mobility is that they "disconnect" from all or part of the net for periods of time, sometimes for milliseconds, sometimes for months/years.

recognising this as a basic feature of internetworking, one would hope that a consistant suite of addressing guidelines would be applicable, regardless of the state of "connectedness".

It's fine to request address space for private networks. It's fine to connect such private networks to some other networks, even one sometimes referred to as the Internet, frequently or infrequently. However, it is not fine to assume you could any form of addressing at all and expect it to be reachable. If you connect to the Internet, you play by its rules, not by the rules of your private networks. -- Pekka Savola "You each name yourselves king, yet the Netcore Oy kingdom bleeds." Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings

| recognising this as a basic feature of internetworking, one would | hope that a consistant suite of addressing guidelines would be | applicable, regardless of the state of "connectedness". This is an important point in my opinion. IPv{46} addresses are a global resource that should be distributed as such. IPv4 adresses are allocated to private networks all the time. Having a connection to the rest is of no concern (and is nowhere stated in any policy to be of any concern). IPv6 addresses will be used for much more than 'just the Internet'. It can be argued that the 2001::/16 prefix SHOULD be used for 'just the Internet', but then still I do not see the point in forcing any kind of connection to one or all other Internet citizens. groet, Pim -- ---------- - - - - -+- - - - - ---------- Pim van Pelt Email: pim@ipng.nl http://www.ipng.nl/ IPv6 Deployment -----------------------------------------------