Search RPD Archives
[rpd] Summary of proposals: IPv4 Runout Management
owen at delong.com
Fri Nov 11 19:53:25 UTC 2016
> On Nov 11, 2016, at 08:31 , Noah <noah at neo.co.tz> wrote:
> On 10 Nov 2016 23:06, "Owen DeLong" <owen at delong.com <mailto:owen at delong.com>> wrote:
> >> On Nov 9, 2016, at 23:14 , Noah <noah at neo.co.tz <mailto:noah at neo.co.tz>> wrote:
> >> If people want to deploy IPv6 they will do it but the compeling reason will eventually be competition as the motivation and nothing else.
> >> Atleast folk i know who do it dont even dual stack in their core as the prefix basically just seats on their boader router facing their ISP for the purpose of announcing it and that is it.
> > Really?
> > You don’t know very many people then… Here are some I know:
> Mine was within the context of AFRINIC region not the rest of the world.
Facebook and Google have substantial deployments in the AfriNIC region.
Care to try again?
I’m betting a lot of networks in the AfriNIC region use Cisco and/or Juniper gear and need access to their web sites.
> > Pinging www.facebook.com <http://www.facebook.com/>:
> > 16 bytes from 2a03:2880:f127:83:face:b00c::25de, icmp_seq=2 hlim=53 time=75.961 ms
> > --- star-mini.c10r.facebook.com <http://star-mini.c10r.facebook.com/> ping6 statistics ---
> > round-trip min/avg/max/std-dev = 75.961/76.957/78.875/1.356 ms
> > Pinging www.google.com <http://www.google.com/>:
> > 16 bytes from 2607:f8b0:4005:801::2004, icmp_seq=2 hlim=55 time=30.146 ms
> > Pinging www.comcast.net <http://www.comcast.net/>:
> > 16 bytes from 2600:1406:34::b819:388d, icmp_seq=2 hlim=56 time=30.565 ms
> > --- a1526.dscg.akamai.net <http://a1526.dscg.akamai.net/> ping6 statistics ---
> > round-trip min/avg/max/std-dev = 30.565/33.799/35.445/2.287 ms
> > Pinging www.netflix.com <http://www.netflix.com/>:
> > 16 bytes from 2620:108:700f::36ba:99e4, icmp_seq=2 hlim=43 time=50.202 ms
> > --- www.latency.prodaa.netflix.com <http://www.latency.prodaa.netflix.com/> ping6 statistics ---
> > round-trip min/avg/max/std-dev = 50.202/50.365/50.512/0.127 ms
> > --- e1280.dscb.akamaiedge.net <http://e1280.dscb.akamaiedge.net/> ping6 statistics ---
> > round-trip min/avg/max/std-dev = 145.721/147.735/149.769/1.653 ms
> > Pinging www.juniper.net <http://www.juniper.net/>:
> > 16 bytes from 2600:1406:1a:3a0::720, icmp_seq=2 hlim=56 time=145.517 ms
> > Pinging www.cisco.com <http://www.cisco.com/>:
> > 16 bytes from 2600:1406:1a:389::90, icmp_seq=2 hlim=56 time=145.636 ms
> I knew about most networks in the USA dual-stacking for a while. But I am less concerned with USA networks.In fact for me, I would be surprised if most if those USA based content providers networks were not running IPv6.
Then you can be surprised. It’s not most (yet), but it is many and becoming more each day.
So if I understand your argument correctly, you are saying that you are not concerned with the state of the rest of the internet, only Africa and you believe that it is OK for Africa to be behind the rest of the world now and that you support policies which will help keep it behind the rest of the world in perpetuity.
I’m sorry to hear that you feel that way. I think such an attitude is counterproductive and will not serve the region well.
> > I know more, but I think that gets the point across.
> It doesn't get any point across from the context of the folks I know within AFRICA. Ours is still very low and there are so many compelling reasons to it. Do you expect Google or Facebook whose businesses are based heavily dependent of numbers to seat there and not roll out v6. In fact they will never turn off their IPv4 because most of the folk outside the USA who consume their services still seat on IPv4 enabled networks. This is the very reason an IPv4 transfer market is booming out at the time where IPv4 is considered legacy.
ROFLMAO… Facebook is already turning off IPv4 everywhere they can.
Cable providers are talking about what it will take for them to be able to stop offering IPv4 services to end users. They’re not talking about “When will end users be happy to give up IPv4”. THey’re talking about “When will it cost us less to lose the few customers that will leave if we turn off IPv4 than it will cost us to maintain the infrastructure necessary to keep IPv4 alive?”
The answer to that latter question is less than a decade, probably less than 5 years at this point.
> >> So announcing an IPv6 prefix to an upstream provider imho doesnt cut it and its easy to do.
> > That’s true… Really deploying IPv6 is what is needed and it is happening with or without the African region.
> I agree and the google stats which I tend to monitor very often attest to that fact but not at the kind of speed one would expect especially in our region with the exception of a new networks.
Correct. But you say this as if it is an acceptable state of affairs for the region and one which should be preserved.
Indeed, you seek to implement policies which will benefit those engaging in this form of bad judgment and poor planning at the expense of those actually deploying useful networks now.
This is the part that I do not understand.
> > Guess what happens in the rest of the world shortly after IPv6 is sufficiently deployed that IPv4 stragglers are no longer considered sufficiently relevant to justify the costs?
> I am not going to guess but its sad that IPv6 deployment will take time as far as our region is concerned especially when the Telecos are not showing signs of immediately firing up v6. We dont have so many content providers out here but even though there is some significant number of startups, most of the video and audio still seats of Youtube and the other mainstream content is yet to go online/Internet whether its Radio or TV content.
Youtube is completely available on IPv6, even in Africa.
> Pretty much most of the content being consumed today on the Internet comes straight from the USA content providers and even when some of their CDN's seat in different locations around the world, the consumers source addresses are heavily IPv4 because ISP's are still happy with what they have.
I know a little bit about the state of things at CDNs and content providers in the US and around the world.
I’m telling you that the days of them continuing to expend resources to support IPv4 eyeballs are, in fact, numbered and not as long as you seem to expect.
As IPv4 continues to get more and more expensive to support, do you really think that international content providers are going to consider it worth maintaining an IPv4 infrastructure just to support the straggling IPv4-only networks within Africa?
Unless you want to see Africa once again left behind as the world moves forward, this attitude of “It’s OK for us to be lagging” has to change.
> > Don’t believe me… Read this:
> > http://www.internetsociety.org/deploy360/resources/case-study-facebook-moving-to-an-ipv6-only-internal-network/ <http://www.internetsociety.org/deploy360/resources/case-study-facebook-moving-to-an-ipv6-only-internal-network/>
> > This is from 2014… They’ve continued to progress down the turn off IPv4 internally road since then.
> > AWS and Microsoft Azure have now deployed IPv6 support for their cloud platforms.
> > Amazon’s ROUTE53 DNS service now has IPv6 support.
> All the above are USA content providers/ISP if you will. Facebook, Google etc makes their money by making sure more people are on facebook or using Google services while they sale their Ads. Neither facebook nor Google or any other content provider network out there care whether the consumers of their services are connecting on the internet through a legacy IPv4 source address if you will or an IPv6 source address. And they are smart and they understand the business which is whey they will obviously not seat back and not deploy IPv6 because every end user matters to them hence their readiness.
Right… Facebook and Google don’t appeal to or attract any users in Africa… Sure. Keep telling yourself that.
Facebook and Google make their money by making sure that their ad revenue exceeds their infrastructure costs. IPv4 infrastructure is becoming more and more expensive while being necessary for a smaller and smaller percentage of users. Eventually there comes a point where the number of users that can’t do IPv6 is small enough that it is no longer cost effective to maintain the infrastructure to support them.
Do you really want to see Africa effectively drop off of social media when that point is reached? That seems to be the argument you are making here.
> > If you truly want to see the African region able to meaningfully interact with the internet going forward, focusing on IPv4 will not serve you well.
> No one if focusing on IPv4, its just that the reality on the ground is exactly what it is.
Yes… The reality is that you seem to think that the following two things are OK and don’t need to be changed:
1. The AfriNIC region is far behind the rest of the world in deploying IPv6.
2. There is little motivation in many of the ISPs in the region to change this fact.
> But even when the IPv4 protocol becomes completely legacy, folk straight outta IPv6 Internet will still need to connect back to the legacy IPv4 network. That is where the IPv4 addresses will still be useful which is why IPv4 transfer markets are cropping everywhere because its about $$$$ at the end of the day.
This statement simply isn’t factual. The transfer market exists because there are people with unused addresses that want to monetize them and people who still need to build out additional IPv4 infrastructure to support end users who cannot yet access content via IPv6.
Once a sufficient critical mass of eyeballs and content are on IPv6, the cost of continuing to support IPv4 will cause it to disappear from the network rather rapidly.
Remember Novell? Remember when the enterprise bought expensive dedicated servers to run Novell networking and paid huge license fees for the server software?
How long after NetBIOS became default on Windows servers before Novell evaporated from the enterprise?
In my experience and memory it took about 3 years, which is virtually overnight in the enterprise world.
If you don’t think that the eyeball providers will start turning off IPv4 compatibility as soon as there’s enough IPv6 content that they don’t have to worry about losing too many customers, you haven’t looked at the economics of it very accurately.
Likewise, if you don’t think that content providers will drop IPv4 like a hot rock as soon as they can do so without disconnecting too much revenue, you are living in an economic delusion and you should get a reality check.
> And that is my rational.
I have a hard time considering something based on economic fantasy rather than reality rational, but to each their own.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the RPD