Re: [Int-area] Still need to know what has changed.... Re: IPv10 draft (was Re: FW: [v6ops] v6ops - New Meeting Session Request for IETF 109 - IPv10)

[Khaled Omar <[e-mail]>]

Hi Fred,

That’s a powerful reply from you both, but let me try to check these statistics for 1 week and do some analysis to know whether I can depend on the result without any kind of forcing the end result, and I feel optimistic that the result will be reflecting the real case, I just need some time to be able to go with a conclusion as I have nothing to check out right now.

As you know from the IPv10 I-D, it requires the network between hosts to be IPv4/IPv6 ready, so it still can be considered, as I still think that the movement to IPv6 still need some time and arrangement to get the best result for the coming generations.

So, let me ask Eric if we can reserve a slot in the next meeting to present IPv10.

Thanks,

Khaled Omar

-----Original Message-----
From: Fred Baker <[e-mail]> 
Sent: Thursday, September 24, 2020 11:45 PM
To: Khaled Omar <[e-mail]>
Cc: Erik Kline <[e-mail]>; Eric Vyncke (evyncke) <evyncke=[e-mail]>; int-area <[e-mail]>; IPv6 Operations <[e-mail]>
Subject: Re: [Int-area] Still need to know what has changed.... Re: IPv10 draft (was Re: FW: [v6ops] v6ops - New Meeting Session Request for IETF 109 - IPv10)

On Sep 21, 2020, at 3:57 AM, Khaled Omar <[e-mail]> wrote:
> Maybe if you can provide me with all the statistics I need that shows the deployment so I can believe.
>  
> Khaled Omar

Sure. I'm using a site that Eric Vyncke has put together and can discuss with you. He uses Google (Erik Kline), Akamai (Jared Mauch), and APNIC (George Michelson/Geoff Huston) numbers; there are other services that publish statistics, he just hasn't included them. As of this instant, Google reports that requests that come to it from 73 countries exceed at least 5% of its workload from that country, and traffic from 37 countries exceed 35% of its workload in that country. Its Eric's site, but the data is from Google, and the site can get you to Akamai and APNIC data as well for the price of a mouse-click.

What I do is download the Google statistics, select the countries that exceed some cut-off, and then ask Eric's or APNIC's site to display the data.

5% cut-off
https://www.vyncke.org/ipv6status/compare.php?metric=p&countries=be,de,in,my,gr,yt,tw,gf,vn,ch,us,fr,mx,pt,jp,lu,br,th,fi,mq,uy,gb,ec,ee,lk,ca,hu,ae,gp,re,nl,tt,ie,au,nz,pe,sa,ga,bo,ro,at,gt,no,ph,cz,sg,il,mo,pl,ar,sx,tg,si,np,mm,om,bt,kr,ke,fo,co,md,zw,cg,pr,is,lv,am,se,ru,li,jo,sk

35% cut-off
https://www.vyncke.org/ipv6status/compare.php?metric=p&countries=be,de,in,my,gr,yt,tw,gf,vn,ch,us,fr,mx,pt,jp,lu,br

APNIC's display of its data on India is interesting https://stats.labs.apnic.net/ipv6/CC?x=1&s=1&p=1&w=30&c=IN

If you scroll down, you will get a break-out by AS. APNIC reports that customers from 12 ASs use IPv6 when accessing APNIC with 50%+ probability ("ipv6-capable" and carrying that amount of data), and given a choice of IPv6 or IPv4, most of them are "ipv6-preferred" (eg, use IPv6 when given a choice). But about 50 ASs actually have users *using* IPv6 for some subset of their workload. In a Financial Times blog a week or two ago, the chair of India's IPv6 deployment task force argued that it should have an IPv6-only DNS Root Server on the basis of its IPv6 deployment and usage. I disagree with him (remarks available on request; they only have 38 IPv6-capable root servers in country), but the basis for the argument was interesting.

I think the APNIC data is interesting because it crosses the backbone. Google and Akamai run CDNs, which means that traffic can be between a residential subscriber and its CDN server without materially touching the ISP. APNIC runs no CDN, which means that traffic has to *also* traverse the ISP and the backbone to APNIC - there is and end-to-end path across the backbone. Think about this: when a user accesses a service using IPv6 (or IPv4 for that matter), the packet has to go from his computer, IOT device, or telephone to the site in question and the response has to come back; there has to be a complete end-to-end path in each direction. Miss one IPv6 connection in one direction, and it may as well be IPv4-only, because that's the only thing the end system will use.

From 73 countries, there is an end-to-end path of sufficient strength that a significant proportion of data *can* traverse it using IPv6, and the end system - which chooses whether to use IPv4 or IPv6 - will *choose* IPv6.

My search engine tells me "There are 195 countries in the world today. This total comprises 193 countries that are member states of the United Nations and 2 countries that are non-member observer states: the Holy See and the State of Palestine." 37% of them, 73, have significant IPv6 usage.

Define "widely deployed"? I'll add "and used?" That's pretty wide, in my book.

What prevents this from being IPv6-only? Computers and network equipment used by residential and enterprise subscribers have supported both IPv4 and IPv6 for years. The most commonly used applications are quite happy with either. The issue I see is primarily enterprise lack of IPv6 adoption in its customer-facing services. Even an "IPv6-preferred" site will use IPv4 when talking with something that will only use IPv4.

There is nothing proprietary here. Forward if you like.

>  From: Fred Baker <[e-mail]>
> Sent: Monday, September 21, 2020 4:49 AM
> To: Khaled Omar <[e-mail]>
> Cc: Erik Kline <[e-mail]>; Eric Vyncke (evyncke) 
> <evyncke=[e-mail]>; int-area <[e-mail]>; 
> [e-mail]
> Subject: Re: [Int-area] Still need to know what has changed.... Re: 
> IPv10 draft (was Re: FW: [v6ops] v6ops - New Meeting Session Request 
> for IETF 109 - IPv10)
>  
> Boy. If “millions and billions” isn’t wide deployment, maybe I need to go back to grammar school.
> 
> Sent from my iPad
> 
> 
> On Sep 19, 2020, at 4:18 PM, Khaled Omar <[e-mail]> wrote:
> 
> 
> But none of these transitioning solutions are widely deployed, maybe 
> it is IPv10 time ;-)
>  
> Khaled Omar
>  
> From: Erik Kline <[e-mail]>
> Sent: Sunday, September 20, 2020 1:05 AM
> To: Khaled Omar <[e-mail]>
> Cc: Fred Baker <[e-mail]>; Eric Vyncke (evyncke) 
> <evyncke=[e-mail]>; int-area <[e-mail]>; 
> [e-mail]
> Subject: Re: [Int-area] Still need to know what has changed.... Re: 
> IPv10 draft (was Re: FW: [v6ops] v6ops - New Meeting Session Request 
> for IETF 109 - IPv10)
>  
> As noted before: RFCs 6052, 6146, 6147, 6877, 7915, and others comprise the solution deployed to literally hundreds of millions if not billions of mobile devices and numerous access networks worldwide.
>  
> On Fri, Sep 18, 2020 at 5:24 AM Khaled Omar <[e-mail]> wrote:
> >> Who are these “many people”, and what problem do they see being solved?
> 
> Network engineers everywhere, they are waiting for the announcement of an official robust solution to the depletion of the IPv4 address space and the division that occurs recently on the Internet.
> 
> People read the draft and many wrote about it because the idea is simple and requires no intervention from their side, that’s why I ask the IETF to take the draft seriously and put personal benefits aside for now, as LATER everything will back to normal, believe me, all are in need for this.
> 
> Khaled Omar
> 
> -----Original Message-----
> From: Fred Baker <[e-mail]>
> Sent: Friday, September 18, 2020 12:24 PM
> To: Khaled Omar <[e-mail]>
> Cc: Roland Bless <[e-mail]>; Eric Vyncke (evyncke) 
> <evyncke=[e-mail]>; int-area <[e-mail]>; 
> [e-mail]
> Subject: Re: [Int-area] Still need to know what has changed.... Re: 
> IPv10 draft (was Re: FW: [v6ops] v6ops - New Meeting Session Request 
> for IETF 109 - IPv10)
> 
> 
> 
> Sent from my iPad
> 
> > On Sep 17, 2020, at 2:08 PM, Khaled Omar <[e-mail]> wrote:
> > 
> > Regarding the confusion, the community is curious about the idea, many people support it as it solves the problem that they think they are not part of it.
> 
> This statement has me a little confused. I see a lot of commentary, 
> but I don’t see people commenting along those lines. I frankly see 
> commentary similar to what I sent you declining a v6ops slot,
> 
> Who are these “many people”, and what problem do they see being solved?
> _______________________________________________
> Int-area mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/int-area