New on RIPE Labs: 1, 000 /22 Allocated from Last /8
[apologies for duplicates] Dear colleagues, We allocated the 1,000th /22 from the last /8. Please read more on RIPE Labs: https://labs.ripe.net/Members/ingrid/1000-slash-22s-allocated-from-last-slas... Kind regards, Mirjam Kuehne RIPE NCC
All, On Friday, 2013-02-01 15:09:58 +0100, Mirjam Kuehne <mir@ripe.net> wrote:
We allocated the 1,000th /22 from the last /8. Please read more on RIPE Labs:
https://labs.ripe.net/Members/ingrid/1000-slash-22s-allocated-from-last-slas...
Just so I understand... It took 140 days to allocate 1000 addresses, or about 7.14 address per day. There are 2^14 /22 in a /8, or 16384. At that burn rate, it will take about 2150 days to finish out the last /8. (16384 - 1000) / 7.14 = 2155 That's about 6 years, assuming things stay constant. 2155 / 365.25 = 5.9 Based on Google's numbers, IPv6 has roughly doubled as a percentage of traffic for the last 3 years... if this continues for the next 6 years, we'll have about 70% of traffic over IPv6 when the RIPE NCC really, REALLY runs out of IPv4 in this region. (Of course, if it continues for 7 years then 140% of traffic will be over IPv6.) ;) It looks like there's likely to be a window of time where new entrants won't be able to get any IPv4 space, and a significant percentage of users will still be IPv4-only. Should we tweak the policy now to make it harder to get IPv4 address space, or wait a few years? It seems slightly unfair to future entrants, but the whole IPv4 allocation model has always vastly favored early entrants, so perhaps we shouldn't worry about it yet. Cheers, -- Shane
Dear Shane,
That's about 6 years, assuming things stay constant. 2155 / 365.25 = 5.9
Many companies in fact had reserves, while they thinked about future. I think the most intresting changes we will see during next 12 month. Amount of requests IPv4 from last /8 can rapidly increase (in 2-5 times). -- Alexey Ivanov LeaderTelecom 04.02.2013 18:02 - Shane Kerr написал(а): All, On Friday, 2013-02-01 15:09:58 +0100, Mirjam Kuehne <mir@ripe.net> wrote:
We allocated the 1,000th /22 from the last /8. Please read more on RIPE Labs:
[1]https://labs.ripe.net/Members/ingrid/1000-slash-22s-allocated-from-last-sla[..]
Just so I understand... It took 140 days to allocate 1000 addresses, or about 7.14 address per day. There are 2^14 /22 in a /8, or 16384. At that burn rate, it will take about 2150 days to finish out the last /8. (16384 - 1000) / 7.14 = 2155 That's about 6 years, assuming things stay constant. 2155 / 365.25 = 5.9 Based on Google's numbers, IPv6 has roughly doubled as a percentage of traffic for the last 3 years... if this continues for the next 6 years, we'll have about 70% of traffic over IPv6 when the RIPE NCC really, REALLY runs out of IPv4 in this region. (Of course, if it continues for 7 years then 140% of traffic will be over IPv6.) ;) It looks like there's likely to be a window of time where new entrants won't be able to get any IPv4 space, and a significant percentage of users will still be IPv4-only. Should we tweak the policy now to make it harder to get IPv4 address space, or wait a few years? It seems slightly unfair to future entrants, but the whole IPv4 allocation model has always vastly favored early entrants, so perhaps we shouldn't worry about it yet. Cheers, -- Shane [1] https://labs.ripe.net/Members/ingrid/1000-slash-22s-allocated-from-last-slas...
Alexey, On Tuesday, 2013-02-05 14:11:35 +0400, LeaderTelecom Ltd. <info@leadertelecom.ru> wrote:
Dear Shane,
That's about 6 years, assuming things stay constant. 2155 / 365.25 = 5.9
Many companies in fact had reserves, while they thinked about future.
But I thought the whole point of the special policy for the last /8 was to support new companies, who did not have a chance to get any reserves? To me, the policy makes no sense otherwise.
I think the most intresting changes we will see during next 12 month. Amount of requests IPv4 from last /8 can rapidly increase (in 2-5 times).
We see a spike in the number of new LIR which happened almost exactly at the same time the IPv4 addresses ran out and the "one allocation per new LIR" policy went into effect: https://labs.ripe.net/statistics/lirs-with-and-without-ipv6 It looks like people are creating new LIR to circumvent the intent of the policy - reserving IPv4 space for newcomers. Cheers, -- Shane
Dear Shane,
But I thought the whole point of the special policy for the last /8 was to support new companies, who did not have a chance to get any reserves?
Correct. I see 2 cases: 1. "Old LIRs". They had some IP-space from previos /8. They had some IPs. And have some networks which were used not very accurate. They can optimize using IPv4 and didn't request IPv4 from last /24. They can give some IPs for clients. But I belive 80% of them will request IPv4 from last /8 during 12 month. 2. "New LIRs". Many "Old LIRs" tell to customers that they can give only 1-4 IPs. So in this case companies which need IPs will request them via RIPE as a new LIRs. I think we should push deployment IPv6 and forget about any improvments policy for IPv4. -- Alexey LeaderTelecom 06.02.2013 16:53 - Shane Kerr написал(а): Alexey, On Tuesday, 2013-02-05 14:11:35 +0400, LeaderTelecom Ltd. <info@leadertelecom.ru> wrote:
Dear Shane,
That's about 6 years, assuming things stay constant. 2155 / 365.25 = 5.9 Many companies in fact had reserves, while they thinked about future.
But I thought the whole point of the special policy for the last /8 was to support new companies, who did not have a chance to get any reserves? To me, the policy makes no sense otherwise.
I think the most intresting changes we will see during next 12 month. Amount of requests IPv4 from last /8 can rapidly increase (in 2-5 times).
We see a spike in the number of new LIR which happened almost exactly at the same time the IPv4 addresses ran out and the "one allocation per new LIR" policy went into effect: [1]https://labs.ripe.net/statistics/lirs-with-and-without-ipv6 It looks like people are creating new LIR to circumvent the intent of the policy - reserving IPv4 space for newcomers. Cheers, -- Shane [1] https://labs.ripe.net/statistics/lirs-with-and-without-ipv6
On 04/02/2013 13:13, Shane Kerr wrote:
All,
On Friday, 2013-02-01 15:09:58 +0100, Mirjam Kuehne <mir@ripe.net> wrote:
We allocated the 1,000th /22 from the last /8. Please read more on RIPE Labs:
https://labs.ripe.net/Members/ingrid/1000-slash-22s-allocated-from-last-slas... Just so I understand...
It took 140 days to allocate 1000 addresses, or about 7.14 address per day.
There are 2^14 /22 in a /8, or 16384.
At that burn rate, it will take about 2150 days to finish out the last /8. (16384 - 1000) / 7.14 = 2155
That's about 6 years, assuming things stay constant. 2155 / 365.25 = 5.9
Based on Google's numbers, IPv6 has roughly doubled as a percentage of traffic for the last 3 years... if this continues for the next 6 years, we'll have about 70% of traffic over IPv6 when the RIPE NCC really, REALLY runs out of IPv4 in this region. (Of course, if it continues for 7 years then 140% of traffic will be over IPv6.) ;)
It looks like there's likely to be a window of time where new entrants won't be able to get any IPv4 space, and a significant percentage of users will still be IPv4-only.
Should we tweak the policy now to make it harder to get IPv4 address space, or wait a few years? It seems slightly unfair to future entrants, but the whole IPv4 allocation model has always vastly favored early entrants, so perhaps we shouldn't worry about it yet.
Deckchairs... deckchairs... Nigel
Nigel, On Tuesday, 2013-02-05 16:21:51 +0000, Nigel Titley <nigel@titley.com> wrote:
Should we tweak the policy now to make it harder to get IPv4 address space, or wait a few years? It seems slightly unfair to future entrants, but the whole IPv4 allocation model has always vastly favored early entrants, so perhaps we shouldn't worry about it yet.
Deckchairs... deckchairs...
Do we care about allowing new entrants or not? If we don't, then we should scrap the last /8 policy, give the entire block out in one big allocation to MyHugeTelco.tld and be done with it. If we do, then we should try to make sure that the policy actually works. I admit to looking at a very short period of time, but it looks like new companies are not going to have any options for IPv4 space in a few years. IPv4 allocation policy tweaks do seem mostly pointless, but they won't be pointless to the start-up founded 4 or 5 years from now, who is trying to connect to IPv4-only users. And who knows, maybe that start-up will be the next Google or Wikipedia, and it will matter to the rest of us too if they succeed or not.... Cheers, -- Shane
* Shane Kerr
On Tuesday, 2013-02-05 16:21:51 +0000, Nigel Titley <nigel@titley.com> wrote:
Should we tweak the policy now to make it harder to get IPv4 address space, or wait a few years? It seems slightly unfair to future entrants, but the whole IPv4 allocation model has always vastly favored early entrants, so perhaps we shouldn't worry about it yet.
Deckchairs... deckchairs...
Do we care about allowing new entrants or not?
If we don't, then we should scrap the last /8 policy, give the entire block out in one big allocation to MyHugeTelco.tld and be done with it.
If we do, then we should try to make sure that the policy actually works. I admit to looking at a very short period of time, but it looks like new companies are not going to have any options for IPv4 space in a few years.
IPv4 allocation policy tweaks do seem mostly pointless, but they won't be pointless to the start-up founded 4 or 5 years from now, who is trying to connect to IPv4-only users. And who knows, maybe that start-up will be the next Google or Wikipedia, and it will matter to the rest of us too if they succeed or not....
I'm not sure I follow the argument here. Wouldn't making it harder to obtain a «last /8» IPv4 block increase the likelihood of your «next Google» failing, because they couldn't obtain the IPv4 space they needed to be successful? -- Tore Anderson
On 6 Feb 2013, at 12:58, Shane Kerr <shane@time-travellers.org> wrote:
Do we care about allowing new entrants or not?
Yeah, but there are limits. I hope/expect our grandchildren won't be discussing 60-70 years from now how future LIRs can qualify for a /32 of v4 or make the remaining RIR reserves ov v4 last beyond the heat death of the universe. :-) At some point v4 simply will run out and no amount of policy tweaking is going to change that. The current trends/policy suggest the NCC has 5-6 years to go. Seems about right. Of course if there's evidence of abuse or LIRs gaming the system, that needs to be dealt with. Do you have data which suggests bad things are happening? FWIW the issue I think we should be discussing is what role, if any, the RIRs have when/if the market in v4 goes mainstream. Debating how we might make the dregs of v4 space last a little bit longer does seem to me to be like rearranging the deckchairs on the Titanic after the iceberg hit.
Debating how we might make the dregs of v4 space last a little bit longer does seem to me to be like rearranging the deckchairs on the Titanic after the iceberg hit.
Having a deckchair or not, might become a topic if you see there is no room left on the lifeboats and you are looking for a floatation device ... presuming that they deckchairs are made from wood and will float .. Or if you want to do the sit and wait game, while everyone else is in panic and jumps. Anyway .. back to the topic .. If we are actually 'care about allowing new entrants', there might be another topic to look into which isn't covered in the current policy for the final /8. And that is the fact that it is allowed to merge LIR's that have been assigned an allocation from the final /8. Currently there are already 6 LIR's to my knowledge that have been merged that now have 2 or more 185.x.y.z /22's ... I've been hearing several discussions in the grapevines from parties that are already low on IP space to use that loop-hole to setup new LIR's, merge them and repeat. To me, I think that we should cut off this route asap, otherwise we might as well hand out the final /8 as if any other /8 .. and be done with it ... Btw.. anyone in need of a deckchair ? Just bring some whiskey and cigars .. let's sit down and watch the show unfold. Regards, Erik Bais
On 2013-02-06 15:32, Erik Bais wrote:
Debating how we might make the dregs of v4 space last a little bit longer does seem to me to be like rearranging the deckchairs on the Titanic after the iceberg hit. Having a deckchair or not, might become a topic if you see there is no room left on the lifeboats and you are looking for a floatation device ... presuming that they deckchairs are made from wood and will float .. Or if you want to do the sit and wait game, while everyone else is in panic and jumps.
Anyway .. back to the topic ..
If we are actually 'care about allowing new entrants', there might be another topic to look into which isn't covered in the current policy for the final /8.
And that is the fact that it is allowed to merge LIR's that have been assigned an allocation from the final /8.
Currently there are already 6 LIR's to my knowledge that have been merged that now have 2 or more 185.x.y.z /22's ... I've been hearing several discussions in the grapevines from parties that are already low on IP space to use that loop-hole to setup new LIR's, merge them and repeat.
To me, I think that we should cut off this route asap, otherwise we might as well hand out the final /8 as if any other /8 .. and be done with it ...
There will be allways such possibility, even if you say that LIRs that are closed shorter than i.e. 5 years after creation have to return addresses - this might be still cheaper to pay 1,8keuro/year than to buy addresses on market. 2keuro + 1.8euro/4 (just one quarter) = 2450 euro for 1024 IP = 2.4 euro per IP. Market price varies from 20$ to 10$ (depending on prefix size, smaller - more expansive). So maybe 5 years blocking period is a good idea as this will be allways money case ? Regards, Marcin
Btw.. anyone in need of a deckchair ? Just bring some whiskey and cigars .. let's sit down and watch the show unfold.
Regards, Erik Bais
On 06/02/2013 14:32, Erik Bais wrote:
To me, I think that we should cut off this route asap, otherwise we might as well hand out the final /8 as if any other /8 .. and be done with it ...
If you're planning to close all of the potential avenue of abuse starting with this one, then let me grab me a chair and some popcorn - I will thoroughly enjoy the hilarious entertainment which will ensue. Nick
On 2/6/13 9:56 AM, Nick Hilliard wrote:
On 06/02/2013 14:32, Erik Bais wrote:
To me, I think that we should cut off this route asap, otherwise we might as well hand out the final /8 as if any other /8 .. and be done with it ...
If you're planning to close all of the potential avenue of abuse starting with this one, then let me grab me a chair and some popcorn - I will thoroughly enjoy the hilarious entertainment which will ensue.
I was just about to say the same thing. Save a chair and some popcorn for me ;) No matter how much policy we do, there is probably no way to outsmart the desperate IPv4-hungry folx in huge need of additional addresses. Cheers, Jan
Hi, On Wed, Feb 06, 2013 at 10:32:13AM -0500, Jan Zorz @ go6.si wrote:
I was just about to say the same thing. Save a chair and some popcorn for me ;)
So how much time do we need for this in Dublin? I'll ask the NCC to provide Popcorn... (No, that was a *joke*, in case someone's sarcasm detector is not turned on). I am listening to the debate, but so far, I more agree with Nick and Jan that people are likely to find loopholes around this, no matter what we do. Tie it to "one LIR one block"? OK, I'll just keep the LIRs around, then. Tie it to "one LIR per person"? OK, I'll find some strawman to be "the LIR admin-c and tech-c". "Make it hideously expensive?" - bad for new entrants, mission failed... So unless someone more creative than I comes up with a good idea what could be done (and finds some backing on the list), I don't think it will be very useful to spend large amounts of precious face-to-face time on this. Gert Doering -- APWG chair -- have you enabled IPv6 on something today...? SpaceNet AG Vorstand: Sebastian v. Bomhard Joseph-Dollinger-Bogen 14 Aufsichtsratsvors.: A. Grundner-Culemann D-80807 Muenchen HRB: 136055 (AG Muenchen) Tel: +49 (89) 32356-444 USt-IdNr.: DE813185279
On 2/6/13 1:38 PM, Gert Doering wrote:
So unless someone more creative than I comes up with a good idea what could be done (and finds some backing on the list), I don't think it will be very useful to spend large amounts of precious face-to-face time on this.
This is like trying to keep the water not rolling down the hill... It will eventually find it's way down anyway, so probably it is better to dig a channels to keep it where you want it ;) Cheers, Jan
On 06/02/2013 18:38, Gert Doering wrote:
So unless someone more creative than I comes up with a good idea what could be done (and finds some backing on the list), I don't think it will be very useful to spend large amounts of precious face-to-face time on this.
I'd be happy to provide popcorn. Anyone got a couch we can all sit on? Nick
On 06/02/2013 20:23, Nick Hilliard wrote:
On 06/02/2013 18:38, Gert Doering wrote:
So unless someone more creative than I comes up with a good idea what could be done (and finds some backing on the list), I don't think it will be very useful to spend large amounts of precious face-to-face time on this.
I'd be happy to provide popcorn. Anyone got a couch we can all sit on?
I have a couch in Dublin, but it's not big enough for the entire WG... We could ask the hotel to import some specially? :) Brian
-----Original Message-----
it is allowed to merge LIR's that have been assigned an allocation from the final /8.
I've been hearing several discussions in the grapevines from parties that are already low on IP space to use that loop-hole to setup new LIR's, merge them and repeat.
What if there are legitimate reasons to merge smaller LIRs? I don't see any way to attack this without creating more costly enforcement issues and needless forms of categorizing and regulating LIRs.
To me, I think that we should cut off this route asap, otherwise we might as well hand out the final /8 as if any other /8 .. and be done with it ...
That is not such a bad option, really. New entrants could acquire them in the secondary market anyway, or from brokers or leases. And if you're a v6 evangelist, the sooner the remaining v4s are vacuumed up, the better, no? But it may be that the status quo is the optimal middle ground. Perhaps 1/4 of the /22s are actually given to the intended type of recipient, it's better than nothing.
Btw.. anyone in need of a deckchair ? Just bring some whiskey and cigars ..
And a brass band! And Kate Winslet!! But why the metaphor of a sinking ship? The v4 internet is not sinking, it's just becoming more efficient.
participants (13)
-
Brian Nisbet
-
Erik Bais
-
Gert Doering
-
Jan Zorz @ go6.si
-
Jim Reid
-
LeaderTelecom Ltd.
-
Marcin Kuczera
-
Milton L Mueller
-
Mirjam Kuehne
-
Nick Hilliard
-
Nigel Titley
-
Shane Kerr
-
Tore Anderson