IPv6
Denis McMahon
RIPE is expected to issue the last of its IPv4 blocks sometime this
summer.
Presumably these will be used up in the not too distant future, and we
will probably see the emergence of hosts who only have IPv6 addresses on
the net by the end of the year.
I wonder if anyone in the UK will be able to connect to those hosts
without using the likes of teredo, freenet6, sixxs etc.
Rgds
Denis McMahon
Tony
>I wonder if anyone in the UK will be able to connect to those hosts
>without using the likes of teredo, freenet6, sixxs etc.
http://www.sixxs.net/faq/connectivity/?faq=native&country=gb
Andrews & Arnold
Bogons
Claranet
Entanet
Goscomb
IDNet
Interoute
provide native IPv6 in the UK I believe.
--
Tony Evans
I'm trying to revive uk.media.films - why not join me there?
[ anything below this line wasn't written by me ]
River Tarnell
I think it's unlikely that connectivity problems will start to appear so
soon, although the costs of providing access to IPv4-only users will
increase rapidly.
For IPv6-only end users, the provider can use CGNAT or proxying to allow
them to connect to IPv4-only sites.
For IPv4-only users, a proxy would also work (this would be basically
"free" for Virgin, for example, because they already use a transparent
proxy for all web access), although I imagine it's easier for most
providers to start offering IPv6 access.
Hosting providers are likely to see the most inconvenience, since they
won't be able to get any more IPv4 addresses, but their customers still
want to provide content to IPv4-only users.
In the short term I imagine they will start moving to IPv6-only networks
internally, and using load balancers with IPv4 addresses to handle
external requests. TLS server name indication can be used to make this
work for TLS sites, although that will require many end-users to upgrade
their browser or OS.
Hosting providers who also have end-users customers (e.g. broadband) can
free a large number of addresses for hosting by moving customers to
CGNAT.
The IPv6 transition has been very badly handled, and is going to cause a
lot more problems than it should have, but I don't think we'll see a
situation where basic web access is impossible.
- river.
Gordon Henderson
Tony <to...@darkstorm.co.uk> wrote:
>In uk.net.providers, Denis McMahon <denis.m....@gmail.com> wrote:
>
>>I wonder if anyone in the UK will be able to connect to those hosts
>>without using the likes of teredo, freenet6, sixxs etc.
>
>http://www.sixxs.net/faq/connectivity/?faq=native&country=gb
>
>Andrews & Arnold
>Bogons
>Claranet
>Entanet
>Goscomb
>IDNet
>Interoute
>
>provide native IPv6 in the UK I believe.
I'm using Entanet's service - Not many IPv6 hosts to talk to
yet...
Gordon
(Gosh - First post I've seen here for years!)
Graham Murray
> I'm using Entanet's service - Not many IPv6 hosts to talk to
> yet...
But topically, the official Royal Wedding site connects using IPv6.
Gordon Henderson
Which one?
The obvious one doesn't appear to...
$ host www.officialroyalwedding2011.org
www.officialroyalwedding2011.org is an alias for appspot.l.google.com.
appspot.l.google.com has address 209.85.143.141
appspot.l.google.com mail is handled by 30 alt3.gmr-smtp-in.l.google.com.
appspot.l.google.com mail is handled by 40 alt4.gmr-smtp-in.l.google.com.
appspot.l.google.com mail is handled by 5 gmr-smtp-in.l.google.com.
appspot.l.google.com mail is handled by 10 alt1.gmr-smtp-in.l.google.com.
appspot.l.google.com mail is handled by 20 alt2.gmr-smtp-in.l.google.com.
Gordon
Matt Balyuzi
> The obvious one doesn't appear to...
>
> $ host www.officialroyalwedding2011.org
> www.officialroyalwedding2011.org is an alias for appspot.l.google.com.
> appspot.l.google.com has address 209.85.143.141
> appspot.l.google.com mail is handled by 30 alt3.gmr-smtp-in.l.google.com.
> appspot.l.google.com mail is handled by 40 alt4.gmr-smtp-in.l.google.com.
> appspot.l.google.com mail is handled by 5 gmr-smtp-in.l.google.com.
> appspot.l.google.com mail is handled by 10 alt1.gmr-smtp-in.l.google.com.
> appspot.l.google.com mail is handled by 20 alt2.gmr-smtp-in.l.google.com.
It appears to be hosted by google, so I suspect the availability of IPv6
depends on your ISP being opted in to IPv6 services from google. AAISP are, so
it works for me -
~: dig @217.169.20.20 AAAA www.officialroyalwedding2011.org
; <<>> DiG 9.8.0 <<>> @217.169.20.20 AAAA www.officialroyalwedding2011.org
; (1 server found)
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 23179
;; flags: qr rd ra; QUERY: 1, ANSWER: 2, AUTHORITY: 0, ADDITIONAL: 0
;; QUESTION SECTION:
;www.officialroyalwedding2011.org. IN AAAA
;; ANSWER SECTION:
www.officialroyalwedding2011.org. 86349 IN CNAME appspot.l.google.com.
appspot.l.google.com. 249 IN AAAA 2a00:1450:8006::8d
;; Query time: 41 msec
;; SERVER: 217.169.20.20#53(217.169.20.20)
;; WHEN: Thu Apr 28 13:11:01 2011
;; MSG SIZE rcvd: 112
--
Matt Balyuzi
Gordon Henderson
OK. I know that Enta were/are in the process of enabling IPv6 with google
- maybe just not yet...
However,
$ ping6 -c5 -q 2a00:1450:8006::8d
PING 2a00:1450:8006::8d(2a00:1450:8006::8d) 56 data bytes
--- 2a00:1450:8006::8d ping statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 4005ms
rtt min/avg/max/mdev = 17.440/18.265/20.664/1.227 ms
So it's routable, just not DNS accessable...
Gordon
Alan J. Wylie
>>But topically, the official Royal Wedding site connects using IPv6.
>
> Which one?
>
> The obvious one doesn't appear to...
>
> $ host www.officialroyalwedding2011.org
> www.officialroyalwedding2011.org is an alias for appspot.l.google.com.
> appspot.l.google.com has address 209.85.143.141
> appspot.l.google.com mail is handled by 30 alt3.gmr-smtp-in.l.google.com.
> appspot.l.google.com mail is handled by 40 alt4.gmr-smtp-in.l.google.com.
> appspot.l.google.com mail is handled by 5 gmr-smtp-in.l.google.com.
> appspot.l.google.com mail is handled by 10 alt1.gmr-smtp-in.l.google.com.
> appspot.l.google.com mail is handled by 20 alt2.gmr-smtp-in.l.google.com.
Google DNS only returns an AAAA record if the query comes from an IPv6
enabled resolver.
http://www.google.com/intl/en/ipv6/
OTOH, Google Mail won't deliver to an IPv6 only MX record.
http://www.google.com/support/forum/p/gmail/thread?tid=1dcc9c70dd4a7225&hl=en
--
Alan J. Wylie http://www.wylie.me.uk/
Andy Davidson
> I suspect that the take-up of IPv6 has been slow because it is so
> massively and incomprehensibly over-specified, instead of merely
> adding an extra group to make the dotted quad into a dotted quin which
> would have been entirely adequate for the foreseeable future, but can
> easily be repeated again in the future if necessary.
I don't see your logic. The v4 address is a fixed 32 bit number. If
we 'grow' it, then it is a a new standard that all devices speaking IP
must be upgraded to, in order to use. So why not lets migrate to the
128 bit address space as offered by IPv6 ?
Additionally, there are significant routing efficiencies offered by v6
that service providers will need to take advantage of, in order to scale
their networks to the likely number of other networks in the future.
(c.f. reducing ~350k v4 pfx routing table to a single slot, or small
number of slots per asn, reduction of >75% of size and complexity.)
> It could even have been backwards compatible if you'd built in an
> assumption to the standard that anything with only 4 0-255s had a
> leading 0 as the fifth (i.e. any 32 bit address is assumed to have 8
> leading zeros to pad it to 40 bit).
I don't share your optimism - you have now proposed variable length
addressing which will be much more computationally expensive for devices
to forward. Also how do 'old world' 32 bit devices talk to the
'new world' 40 bit devices ? Or the other direction ?
> Existing routers would have been far more likely to have been able to
> cope with this by means of a firmware update, instead of hardware
> replacement (although that's probably what the major players with
> input to the standards wanted - something that would force their
> customers to have paid-for upgrades instead of free software patches).
Again, you are confused, it has been possible to build a v6 native
network since before 2006, why would any network operator have bought
any equipment that can not just be firmware upgraded to v6 support for
the last five years or more ?
> People were probably hoping that something at least vaguely resembling
> sense would prevail, and something usable would be produced, with
> numbers that it's possible for a human being to remember for at least
> long enough to write them down.
I can remember v6 addresses, they are easier than v4, because you can
map your network topology into the address. You can't do this simply
with v4.
> There was never any need to produce a system that has enough addresses
> to allow for the most optimistic of all human expansion until the heat
> death of the universe.
But there was a need for a system that could cope with future network
scaling issues with sparse addressing.
You are thinking in terms of 2010 network design and applications,
whereas v6 is designed to power the applications and networks of
the future.
You also should consider that only people who know what an IPv4
address is, will need to know what an IPv6 address is, and that end
users will not need to be concerned by the new technology. IPv6
really does 'just work', and is hidden to the end users. If an
engineer really doesn't have the first clue about v6 -- even after
the hugely successful World IPv6 Day two weeks ago -- then I am
worried that this engineer is already obsolete.
Andy