On 13/02/2007 9:40 p.m., Jamie Baddeley wrote:

*characterised by entities who've trod the multihoming path to APNIC to get *assigned*, plumped up their estimates, and is still looking at at the remaining space wondering what the heck they can do with it. There is a minimum portable allocation / assignment - it would take an APNIC (or is it the laws of physics [routing], Jim?) policy change to alter that and get numbers back for reallocation / assignment.

Gerard

On 13 Feb 2007, at 20:25, Don Stokes wrote:

Frankly, I don't think the size of the routing table is an excuse any more. Memory and processor horsepower are cheap.

It's not just the extra memory and grunt needed, convergence time needs to stay low too.

Hi all, I thought I'd jump in here. Without attempting to stop any discussion here on the nznog list, I would like to invite you all to continue this discussion on the policy-sig list! This is a quite significant proposal that raises some important questions about the way IP space is managed and how we as the Internet community should manage the eventual exhaustion of IPv4 and increased deployment of IPv6. As such, it affects all of us in the Internet community. The nznog list has always had very healthy debate and although some of us from APNIC staff are on this list and try to capture some of the feedback here, the way to really get your voice heard is to do it on the policy-sig list. (I see that some of you have done that which is great, but I would like more "nznogers" to get on the policy list and get involved!) There are a lot of valuable opinions here and this discussion needs input from all parts of the community. You can easily subscribe to the policy-sig list here: http://mailman.apnic.net/mailman/listinfo/sig-policy And check out the archives here: http://www.apnic.net/mailing-lists/sig-policy/ Don't hesitate to shoot me a mail if you have any questions about this! Cheers, Nurani APNIC P.S. I was very sad to miss the last NZNOG meeting! But I did have a nice cold beer here and I thought of you guys... :-) On 15/02/2007, at 6:06 AM, jamie baddeley wrote:

On Wed, 2007-02-14 at 09:25 +1300, Don Stokes wrote:

...

Jamie Baddeley wrote: > IX's and the like have accelerated IPv4 depletion. Discuss. Only true to a limited extent. The real problem is that the minimum allocation sizes have been set artificially high to keep the size of the routing table down. If it had been possible to get /24s from the RIRs in recent years, I'd say a lot of multihomed sites wouldn't be sitting on large blocks of unused address space. I see the many more /24s finding their way into the routing table as organisations acquire unused blocks from others. It should have been possible a long time ago, but RIR rules have discouraged it. Like it or not, once the RIRs have no address space left to assign, a secondary market in address space will form, and it's up to the RIRs to figure out whether they're going to be involved in the process or not.

Indeed. And the question has been asked: http://www.apnic.net/mailing-lists/sig-policy/archive/2007/02/ msg00007.html

jamie _______________________________________________ NZNOG mailing list NZNOG(a)list.waikato.ac.nz http://list.waikato.ac.nz/mailman/listinfo/nznog

On Wed, Feb 14, 2007 at 10:41:54AM +1300, Alastair Johnson said:

Jamie Baddeley wrote:

...

IX's and the like have accelerated IPv4 depletion. Discuss.

-Multihoming means assignment at the cost of conservation. -Has also accelerated AS number depletion (though that issue seems to have been sidestepped.)

I don't think the IXes are to blame at all. General end organisation multi-homing is far more to blame than peering. How many people are happily peering at WIX using PA space and private ASNs and having the RS' strip them?

Assuming this wasn't a rhetorical question, quick greppage of the looking glass output suggests 55 peers have private ASN and are advertising a non-zero number of prefixes. I'm not sure this supports an arguement of "peering causes v4 exhaustion" - I'd hypothesize that the majority of NZ organisations going to APNIC for a multihoming allocation aren't peering (but I've no data to back that assertion up). It would be worthwhile noting here that our policies for APE and WIX vary quite significantly here - on WIX we allow /29 and private ASN, on APE we tend to try and enforce /24 and public AS. Generally, if anybody comes to us and starts making muttering noises about wanting to multihome with both an IX and a transit provider, we push them towards a public ASN - it gets too hard to do sane path stuffing if the IX/ISP is stripping the private ASN. Apologies for the late thread wade in, I've been on holiday. Cheers Si

From: Robert Gray [mailto:bob(a)brockhurst.co.nz] What if a large country with a huge population, an independent outlook on the world and vast growth in internet usage (mentioning no names) was denied more allocations and simply decided to recycle someone else's /8's as there's "nothing our people need to see in there".

Ok, I'll bite: 16/8 They don't exist any more 19/8 They could sell the block to help get out of debt 20/8 Sorry, think my NDA still applies 34/8 We don't like them 38/8, 40/8, 47/8, 48/8, 52/8, 53/8, 54/8 They don't need/fully utilize the space That should give us a couple more years, eh? JB

On 14-Feb-2007, at 13:35, Robert Gray wrote:

Jamie Baddeley wrote:

...

Given the deafening silence I'm going to provoke some debate. Just for fun :-)

Debate on this subject is very important as the policy proposal suggests "the exhaustion of IPv4 address space is expected to take place as early as within the next five years".

Imagine a world where there are no more allocations of v4 space. How would this affect your business and customers?

There's a slight non-sequitur in your phrasing, there. If the RIRs have no more address space to assign or allocate, that doesn't mean there are necessarily "no more allocations of v4 space"; it might just mean that you stop shopping for address space at APNIC, and start buying on eBay.

What if a large country with a huge population, an independent outlook on the world and vast growth in internet usage (mentioning no names) was denied more allocations and simply decided to recycle someone else's /8's as there's "nothing our people need to see in there".

Such large countries are often responsible for prodigious international trade; it's not obvious that disabling lines of communication would benefit that trade, so perhaps that would be a disincentive to such practices.

What if this was 202.0.0.0/8 or 203.0.0.0/8, or both. Could that happen? If so how badly do things get broken?

There's a difference between internal-use addressing and globally- unique addressing. If a large carrier decided to claim something like 202/8 or 203/8 for their own and attempted to announce them (in whatever small chunks were required to win their traffic) we might reasonably expect their peers and transit providers not to accept and propagate the announcements. That limits the usefulness of such hijacking.

Seems to me doing nothing (in this case not even rearranging the deck chairs) is not a good option but it looks like the one that will be taken unless there is agreement on a better plan. This policy proposal needs some more debate and who better that nznog members to 'stir it up' :)

http://www.apnic.net/mailing-lists/sig-policy/index.shtml

The cynic in me (hey! that's all of me!) says that IPv6 is the answer, but that until the cost of deploying IPv6 exceeds the cost of struggling on with IPv4, everything will be business as usual. This further suggests to me that we have at least ten years of intense lawsuits and shouting to go before I can go IPv6-only. Joe