Re: [uknof] test website with an IPv6 address

[Tim Bray]

On 27/05/2020 17:28, Greg Choules wrote:

Good evening all.
I am looking for a simple web page that can be reached over IPv6, for 
use by an intermittent test tool. There are a couple of caveats:


 1. It must be HTTP only with no redirect to HTTPS (you'll see why).
 2. It must not care about the domain used in the GET. This is because
I want to configure my own  records with specific TTLs, for
the DNS element of the test tool.



http://ip6.provu.co.uk - returns nothing but the IP address you came 
from.   2001:41c9:2:d2::40:1



Just moved onto a dedicated IP - for some reason it wasn't, when I have 
a zillion IPs to go at.



Presume you aren't going to do more than a 200k hits a day?


See also

http://ip.provu.co.uk/ (iframe that give IPv4, IPv6 and IP whatever)


and

http://ip4.provu.co.uk/    - not currently on a dedicated IP.

http://ipa.provu.co.uk


(and thanks that my Three phone sometimes has IPv6, and sometimes doesn't)




--
Tim Bray
Huddersfield, GB
t...@kooky.org
+44 7966479015

Re: [uknof] test website with an IPv6 address

[Giles Coochey]

Is this for a constant monitoring process you want?


If so, can think of a few reasons why picking on a website might not be 
such a good idea, if you can't think of one right off the back of your hand.



If it is for some occasional testing type thing then I have some things 
that might fit your bill (home testing type thing).



On 27/05/2020 17:28, Greg Choules wrote:

Good evening all.
I am looking for a simple web page that can be reached over IPv6, for 
use by an intermittent test tool. There are a couple of caveats:


 1. It must be HTTP only with no redirect to HTTPS (you'll see why).
 2. It must not care about the domain used in the GET. This is because
I want to configure my own  records with specific TTLs, for
the DNS element of the test tool.

Any takers?

thanks, Greg


Greg Choules
Design Engineer
07957 807004
www.three.co.uk 
*We’re building the UK’s fastest 5G network*

This e-mail message (including any attachment) is intended only for 
the personal use of the recipient(s) named above. This message is 
confidential and may be legally privileged. If you are not an intended 
recipient, you may not review, copy or distribute this message. If you 
have received this communication in error, please notify us 
immediately by e-mail and delete the original message.


Any views or opinions expressed in this message are those of the 
author only. Furthermore, this message (including any attachment) does 
not create any legally binding rights or obligations whatsoever, which 
may only be created by the exchange of hard copy documents signed by a 
duly authorised representative of Hutchison 3G UK Limited. Hutchison 
3G UK Limited is a company registered in England and Wales with 
company number 3885486. Registered Office Star House, 20 Grenfell 
Road, Maidenhead, Berkshire SL6 1EH


--
Giles Coochey

[uknof] test website with an IPv6 address

[Greg Choules]

Good evening all.
I am looking for a simple web page that can be reached over IPv6, for use by an 
intermittent test tool. There are a couple of caveats:

  1.  It must be HTTP only with no redirect to HTTPS (you'll see why).
  2.  It must not care about the domain used in the GET. This is because I want 
to configure my own  records with specific TTLs, for the DNS element of the 
test tool.

Any takers?

thanks, Greg

[cid:799f51b8-16a2-4cfb-8f48-e369f4a12b19]
Greg Choules
Design Engineer
07957 807004
www.three.co.uk
We’re building the UK’s fastest 5G network

This e-mail message (including any attachment) is intended only for the 
personal use of the recipient(s) named above. This message is confidential and 
may be legally privileged.  If you are not an intended recipient, you may not 
review, copy or distribute this message.
  If you have received this communication in error, please notify 
us immediately by e-mail and delete the original message.

Any views or opinions expressed in this message are those of the author only. 
Furthermore, this message (including any attachment) does not create any 
legally binding rights or obligations whatsoever, which may only be created by 
the exchange of hard copy documents signed by a duly authorised representative 
of Hutchison 3G UK Limited. Hutchison 3G UK Limited is a company registered in 
England and Wales with company number 3885486.  Registered Office Star House, 
20 Grenfell Road, Maidenhead, Berkshire SL6 1EH

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Per Bilse]

 I'm saying that address space is a limited resource, and as such it cannot be 
expected to be free.  Cheap, maybe, but one way or another it will always cost 
something; even staking a free claim isn't free.  This is a general principle 
that applies universally, and I recall Daniel's note as a slightly 
tongue-in-cheek, gentle reminder that reality prevails, even in the Internet.
Running out of IPv6 space was mentioned by Daniel in connection with somebody 
having asked why people couldn't just get the IPv6 space they wanted, without 
any rules/guidelines/approval; it's virtually unlimited, isn't it?  The issue 
was crude landgrab attempts that were rejected, not legitimate, bona fide 
requests.  It's another example of needing demand-sensitive pricing or rules 
and enforcement; in the absence of both, somebody will always try to move the 
(hexa)decimal point.
  -- Per
On Tuesday, 26 May 2020, 18:51:04 BST, Leo Vegoda  wrote:  
 
 On Tue, May 26, 2020 at 8:02 AM Per Bilse  wrote:
>
> Daniel Karrenberg (founder/CTO RIPE NCC) has for decades said "I sell IP 
> address space".  The price has historically been low, but it has always been 
> a finite resource, and that goes for IPv6 address space too (when the RIPE 
> NCC opened up for IPv6 address space, it took half an hour to receive enough 
> requests to exhaust the entire IPv6 address space.)

I'm not really sure what you are trying to say. But I don't think the
RIPE NCC has ever had so many requests for IPv6 that it would take all
the IPv6 space for the very simple reason that there isn't enough
network to justify it.

The RIRs collectively took a couple of years to get 100 requests for
IPv6 /35s, back in the day. I remember preparing the announcement that
we had hit that policy milestone.

Now, there is a settled policy for IPv6 allocations to ISPs and each
of the RIRs get a /12 allocation from the IANA from a /3. Those /12s
last a long time and there are enough /12s in that /3 that even if
each RIR needed one a year it would take a century to allocate them
all.  

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Tim Bray]

On 26/05/2020 22:59, Pete Stevens wrote:


I wonder what the absolute minimum set if before $ultra-cheap-broken-isp
just goes ipv6+nat64 and doesn't care about breaking other stuff. Free 
broadband that comes with your mobile contract / cornflakes could be a 
candidate. 



I'd suggest the big porn sites might need IPv6 for that to float.



Slightly different scenario for sweeping up sites to have IPv6 on server 
side:


It might one day that 464xlat fails at $massive_mobile_isp - most 
punters still happy because facebook and google still up.

And the headline reads `twitter down on $massive_mobile_isp'

And people phone $small_supplier and say `Can't get on your website from 
my mobile, and its your fault because google and facebook still work`




--
Tim Bray
Huddersfield, GB
t...@kooky.org
+44 7966479015

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Leo Vegoda]

On Wed, May 27, 2020 at 2:22 AM Per Bilse  wrote:
>
> I'm saying that address space is a limited resource, and as such it cannot be 
> expected to be free.  Cheap, maybe, but one way or another it will always 
> cost something; even staking a free claim isn't free.  This is a general 
> principle that applies universally, and I recall Daniel's note as a slightly 
> tongue-in-cheek, gentle reminder that reality prevails, even in the Internet.

The address space itself is free. That's why you can use as many
probabilistically unique /48s in fd00::/8 - you just can't register
them. You're paying for the registration, the DNS, and a bunch of
other services, like RIPE meetings, that are associated with the
registration.

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Simon Lockhart]

On Wed May 27, 2020 at 12:27:49pm +0100, Paul Mansfield wrote:
> according to the SOA for bbc.net.uk there should be a b...@bbc.co.uk
> who can answer why!

Because I don't remember writing IPv6 support in my (perl) load balancing DNS
server :)

BBC have, however, run IPv6 on www.bbc.co.uk at periods in the past (e.g. IPv6
day).

Simon
(bofh@bbc, at least at some point in the past)

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Robert McKay]

On 2020-05-26 15:13, Paul Mansfield wrote:

Meanwhile, are there any ISPs who run a DNS64/NAT64 as a service? It
strikes me that it would be a good exercise/practise.



go6lab runs a public DNS64/NAT64 service.. just change your nameserver 
to their IPv6 nameserver and enjoy free access to the legacy internet. 
They synthesise  records pointing at their network. No account 
required. Of course, no SLA either.


https://go6lab.si/current-ipv6-tests/nat64dns64-public-test/

Rob

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Jacob Mansfield]

> I view the v6 transition like the how to get rid of IE6 transition.

This seems like a fairly good comparison. Just like with IE, we're stuck in
a catch-22 until something blows IPv4 out of the water.
There's no incentive for making consumer devices support IPv6 until ISPs
stop supporting v4, and that's very unlikely to happen unless there are
IPv6-only networks that the devices won't work on.
The fundamental problem is that there's no business case on either end,
which means it all comes down to risk. Why should either bother to add new
functionality when as far as they're concerned nothing is broken and
there's no profit to be made from upgrading.

Unfortunately the same applies to the majority of consumers, who
realistically don't care how their internet works as long as they can
access Facebook/Candy Crush/. Which ultimately means at
the moment there's nobody to pressure any party into making the upgrade.
Whilst Paul's suggestion may not be ideal, it feels more likely to succeed
than the "sit and wait" of the past 24 years.

Regards,
Jacob Mansfield
www.jacobmansfield.co.uk

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Christian]

Lots of chickens and eggs cycles to resolve. What is not clear to me is 
whether today people running hosts with no or incomplete IPv6  support  
know IPv6 is something they should be on top of. That at least would be 
progress. As for many years they were being told by their ISPs / Vendors 
/ Consultants / Experts that IPv6 is not a thing they should be  
planning let alone implementing.


If largely aware. Is there a touch of complacency at hosts that they can 
configure up IPv6 quickly when they need to?  For instance by throwing 
some money via a CDN?


The local chapter has had a long running IPv6  host monitoring project 
ipv6matrix.org  which we've been hosting at University of Southampton 
that looks at the top million domains as set by Alexa. Olivier still  
keeps it regularly updated.


It's got a d3 graphical charting displays but for open data junkies the 
data is there for direct use too.



Christian


On 27/05/2020 11:52, Nick Hilliard wrote:

Paul Mansfield wrote on 27/05/2020 11:47:
I was surprised how many services aren't but you'd think they 
could/should be

https://ipv6.watch/


this should give some indication of the complexity, and therefore the 
cost, of service availability over ipv6.


Nick
ink

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Nick Hilliard]

Will Hargrave wrote on 27/05/2020 12:46:
I’m sure you know this but: what this misses is the vast amount of their 
actual CDN traffic, i.e. the actual bulk of the content. I don’t think 
i’m giving away secrets when I say there is substantial IPv6 traffic there.


Most people will never email Netflix and barely look at their website.


this is exactly the point though.  Cherry-picking the higher volume data 
sources and sinks is one thing, but the real cost and diminishing value 
of ipv6 deployment is associated with the long tail - both from a 
content publisher and content consumer point of view.


Nick

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Will Hargrave]

On 27 May 2020, at 12:28, Pete Stevens wrote:

That is quite interesting, but could really do with some information 
about what improvements are needed. Netflix is listed,
but I've got no clue what they are expected to do to improve their 
v6 connectivity.

The first fails our checker
https://www.mythic-beasts.com/ipv6/health-check?domain=netflix.com=#test-details-mailserver-reverse-dns
so Netflix get 10/11.


I’m sure you know this but: what this misses is the vast amount of 
their actual CDN traffic, i.e. the actual bulk of the content. I don’t 
think i’m giving away secrets when I say there is substantial IPv6 
traffic there.


Most people will never email Netflix and barely look at their website.

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Paul Mansfield]

On Tue, 26 May 2020 at 17:12, Brandon Butterworth  wrote:
> One would be regulation in some manner, eg it becomes part

Ofcom last mentioned ipv6 in 2017 as far as I can tell
https://www.ofcom.org.uk/search?query=ipv6=

> So to answer original question - set up a process that will
> result in a date you can announce.

maybe the Gov't gigabit/fibre voucher scheme?
https://gigabitvoucher.culture.gov.uk/

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Pete Stevens]

That is quite interesting, but could really do with some information about what 
improvements are needed. Netflix is listed,
but I've got no clue what they are expected to do to improve their v6 
connectivity.


Netflix use Google for email, and Google are missing some reverse DNS on
their IPv6 addresses.

e.g. from Mythic I see

$ dig aspmx3.googlemail.com  +short
2404:6800:4003:c06::1a
$ dig -x 2404:6800:4003:c06::1a +short
$

from IDNet I see
$ dig aspmx3.googlemail.com  +short
2a00:1450:4010:c03::1a
$ dig -x 2a00:1450:4010:c03::1a +short
lr-in-x1a.1e100.net.
$ dig lr-in-x1a.1e100.net  +short
2a00:1450:4010:c03::1a


The first fails our checker

https://www.mythic-beasts.com/ipv6/health-check?domain=netflix.com=#test-details-mailserver-reverse-dns

so Netflix get 10/11.

Pete

--
Pete Stevens
p...@ex-parrot.com
http://www.ex-parrot.com/~pete/
https://www.mythic-beasts.com/
https://twitter.com/Mythic_Beasts

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Paul Mansfield]

ah, bbc.co.uk does have IPv6 addresses

$ dig +short  bbc.co.uk
2a04:4e42:200::81
2a04:4e42:600::81
2a04:4e42:400::81
2a04:4e42::81

but not www.bbc.co.uk which is CNAME to bbc.net.uk

according to the SOA for bbc.net.uk there should be a b...@bbc.co.uk
who can answer why!

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Paul Mansfield]

On Wed, 27 May 2020 at 11:56, Paul Bone  wrote:
> C:\Users\paul.bone>nslookup www.bbc.co.uk

I seem to recall the BBC did have a v6 enabled front end at one point.
presumably the licence fee needs to increase before they can afford to
run both ;-)

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Paul Mansfield]

On Wed, 27 May 2020 at 11:56, Dave Bell  wrote:
>> https://ipv6.watch/
>
> That is quite interesting, but could really do with some information about 
> what improvements are needed. Netflix is listed, but I've got no clue what 
> they are expected to do to improve their v6 connectivity.

agreed. I followed the github link, so I guess you can spin up a
version with more diagnostics?
https://github.com/andir/ipv6.watch

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Paul Bone]

This is one that has always surprised me:

C:\Users\paul.bone>nslookup www.bbc.co.uk
Server:  UnKnown
Address:  10.234.120.1

Non-authoritative answer:
Name:www.bbc.net.uk
Addresses:  212.58.237.254
  212.58.233.254
Aliases:  www.bbc.co.uk



On Wed, 27 May 2020 at 11:54, Paul Mansfield 
wrote:

> On Tue, 26 May 2020 at 22:59, Pete Stevens  wrote:
> > A v6 only end user ISP can already access anything behind cloudflare,
> > facebook, google, youtube, netflix but not twitter.
>
> I was surprised how many services aren't but you'd think they could/should
> be
> https://ipv6.watch/
>
>

-- 
Paul Bone
Network Consultant

PMB Technology

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Dave Bell]

On Wed, 27 May 2020 at 11:49, Paul Mansfield 
wrote:

> I was surprised how many services aren't but you'd think they could/should
> be
> https://ipv6.watch/


That is quite interesting, but could really do with some information about
what improvements are needed. Netflix is listed, but I've got no clue what
they are expected to do to improve their v6 connectivity.

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Nick Hilliard]

Paul Mansfield wrote on 27/05/2020 11:47:

I was surprised how many services aren't but you'd think they could/should be
https://ipv6.watch/


this should give some indication of the complexity, and therefore the 
cost, of service availability over ipv6.


Nick

Re: [uknof] 9pm data dip - update

[Greg Choules]

Good morning all.
Now that Ramadan has ended and we have a few days' worth of stats to look at it 
does indeed appear that the dip was related. There was one on 23rd (Saturday) 
but none since. Throughout this period it was shifting to the right each day, 
broadly tracking sunset in the UK.

Thank you for the discussion this prompted.

cheers, Greg

From: Gord Slater 
Sent: 23 May 2020 16:13
To: Paul Mansfield 
Cc: uknof@lists.uknof.org.uk ; Greg Choules 

Subject: Re: [uknof] 9pm data dip

Though beware that "it's complicated" :) If you compare data against majority 
muslim countries, you'll see some Ramadan effects dragging on a little for the 
next month or so. I doubt this will be noticeable in UK data though, except in 
certain local geo areas.


Eid is broadly equivalent to Christmas in scale, so the festivities and family 
celebrations tend to last at least couple or three days. similar to the western 
Christian tradition of eating turkey sarnies until leftovers are used up. The 
first day is usually the most significant.
After Eid settles down, there are broadly 3 groups of behaviour:-


Extra optional religious fasting - this is approximately a week or more of 
additional fasting for religious purposes, or replacement fasting for people 
who could not fast in Ramadan for travel or temporary health reasons.

Extra lifestyle fasting - some people (possibly as many as 5 to 10% in some 
societies) fast on and off over the following month or two because they enjoy 
it but have no significant religious intent. Some do it to provide moral 
support for friends or  amily who fast for more othodox reasons.

"back to normal" - where people carry on their normal lives - this is the 
majority of the working populace in majority-muslim countries, especially doing 
manual labour


Also significant when comparing international data is that majority-muslim 
countries keep very different business hours during Ramadan, which will show up 
in data plots.  The data I've seen in the past was more akin to Spanish-style 
summer siesta pattens (most govenrment offices and large public-facing 
businesses are only open in the mornings during Ramadan, some are closed all 
day Fridays)

Also the normal business "weekend" varies a lot across the muslim world - some 
places are closed Thursdays and Fridays, (in West Africa this can vary from 
state to state within countries) others close Saturday an Sunday with an 
afternoon break on Friday, others close Friday and Saturday,
Companies and governement departments dealing with non-muslim countries may 
keep two sets of hours, the "international business" hours and their normal  
customer-facing hours

The cultural concept of time can be very different from the western perception 
of opening times. Priorities and approaches to opening times and appointments 
are heavily influenced by cultural situation, especially in very hot countries.

Personally, I'm betting on the "hey kids get to bed it's getting dark" theory, 
combined with Covid-19 lockdown effects. Mainly because - as I'm try to explain 
in the shortest possible manner (but failing badly) - comparisons between UK 
and majority-muslim countries will be extremely difficult.
As a rough guess I'd say that Pakistan and Bangladesh would be the closest fit 
to cultural practices durind Ramadan, but international family communications 
UK<>Asia/ISC across the timezones would skew the data significantly. I also 
have less data/comms insight there than in MENA, so I'm not really sure.

There may however be some close correlation between French ISPs, Dutch and 
Belgian ISPs and their Maghrebi ISP counterparts, being largely in the same 
timezone, though the UK only has a very small Magrebi diaspora compared to any 
of those other 3 countries. Marseilles would be a good geo-area to compare to 
Tunisia, Algerian and to a lesser extent, Moroccan data plots for example, 
given similar sunrise/sunset times.




On Sat, 23 May 2020 at 10:54, Paul Mansfield 
mailto:paul%2buk...@mansfield.co.uk>> wrote:
Hopefully Greg can tell us if the bandwidth dips go away now it's the
last day of Ramadan, as that seems to be the popular explanation.
The first day of Ramadan was 23rd April so the dips should have started then.



--
sent via Gmail web interface, so please excuse my gross neglect of Netiquette

This e-mail message (including any attachment) is intended only for the 
personal use of the recipient(s) named above. This message is confidential and 
may be legally privileged.  If you are not an intended recipient, you may not 
review, copy or distribute this message.
  If you have received this communication in error, please notify 
us immediately by e-mail and delete the original message.

Any views or opinions expressed in this message are those of the author only. 
Furthermore, this message (including any attachment) does not create any 
legally binding 

Re: [uknof] Thought for the day: announce the end of IPv4 internet connections by 2026

[Jonathan McDowell]

On Tue, May 26, 2020 at 10:59:28PM +0100, Pete Stevens wrote:
> > Unfortunately the same applies to the majority of consumers, who
> > realistically don't care how their internet works as long as they
> > can access Facebook/Candy Crush/.
> 
> A v6 only end user ISP can already access anything behind cloudflare,
> facebook, google, youtube, netflix but not twitter.

I broke v4 on my wired lan last week (rogue DHCP server on a development
OpenWRT box) and it was 15 hours before I noticed (admittedly part of
that I was asleep). It was Twitter that made me investigate, everything
else (ssh, email, whatever other websites I was looking at) was all
fine.

Last time I ran numbers (about 6 months ago) 50% of my home traffic went
over v6 (without any attempt to coerce traffic to prefer it over v4).

J.

-- 
Revd Jonathan McDowell, ULC | I'm only here for the salad bar.