IPv6 Trial Update

[#1]

Thanks jab1,
anyway will be happy if someone of plusnet answer my question.

[#2]

When @dave is back, no doubt he will.

[#3]

Not from PlusNet, but it is a /56 global address.

[#4]

Thanks IMM, that's sound really nice.

[#5]

If it helps my IPv6 IP starts with  2a02:

IPv6 PPPoE is set to share the IPv4 session.

[#6]

I can't speak for the future roll out but the current trial is deploying GUA to the customer (/64), with prefix delegation of the associated /56 available.  I believe it is static - at least mine hasn't changed throughout the trial.

There is nothing stopping anyone using ULA at home in addition, but NAT66 is ugly and not the recommended deployment method for IP6.  Plusnet would have to comment on what the home hub's will support (but the service doesn't require use of Plusnet's router).  You can't just use ULAs as they wouldn't route beyond your home network.

LL addresses are just part of the standards and depend mostly on the end points more than anything else.  ( Linux can be configured to auto-generate them or not, for example. ). They're defined for use between endpoints on the same link-layer network and don't cross subnet boundaries if you separate up wifi from lan, IoT devices, guest/guest network, home lab, etc.

IP6's design ethos is multiple concurrent addresses can be assigned on an interface.  Quite different to IP4+NAT.

The trial retained the static IP4, and is running dual-stacked, PPPoE.

Hope that helps, though as I said, it's what's provided in the trial and might not be what is done in the full release.

Cheers,

Mark

[#7]

Mark thanks for the information. I'm Really interesting in GUA, but more concerned to keep my IP4 also.
Hope in the future has more slots to join.
Regards

[#8]

Quoting MPC: "mine hasn't changed throughout the trial."

Mine has, several times. My connection has dropped every 10-14 days for a brief period overnight, and on each occasion the IPv6 address assigned to my router has changed.

[#9]

Interesting, Jaggies.  I haven't had any significant disruption so perhaps that's why I've not seen a change.

I like the resilience of key services at home being static, and the simplicity of using the GUA directly on a couple of things.

I also have some externally accessible IP6 DNS entries pointing home.  DNS caching being what it is, there's a service mis-direction possibility if the prefix were to change just after being cached by an external mail server.

Were Plusnet to issue 'my' prefix to someone else quickly with respect to that cache time, my email might end up arriving at someone else's server.  Something I've not had to worry about for IP4 with the static mapping.  It would be a shame if that wasn't true for IP6 as well.

Cheers,

Mark

[#10]

---

@dxuk wrote:
In case of joining trail, are you removing static IP4 or stacking ipv6 to the existing?

---

Those of us on the high-touch network trail have dual stacks our static IPv4 and a IPv6 service.

EDIT: I should have read all of the subsequent responses before replying to a post on the previous page!!

[#11]

Hi,

The high touch trial which everyone posting is on is static IPv4 and static IPv6, fully dual stack with all bar one person having a /56 prefix. The low touch trial, of which I only have my test line will be dynamic IPv4 and dynamic IPv6, IPv6 should be fairly sticky though but could change.

@Jaggies the IPv6 prefix should stay the same but it's normal for the /128 to change on devices. My Mac had about 8 different addresses last time I looked and the IP it was using seemed to be different every day. My Linux server has a fixed IP as it's running a web server.

[#12]

---

@dave wrote:

The low touch trial, of which I only have my test line will be dynamic IPv4 and dynamic IPv6

---

Thanks for the update Dave.

Will this be the shape for the eventual rollout, with dynamic IPv4 users getting dynamic IPv6 as well?

Will there be a wider (volunteer-able) low touch trial at some point, or are you intending to go directly to rollout from the current trial lines?

[#13]

---

@Jaggies the IPv6 prefix should stay the same but it's normal for the /128 to change on devices. My Mac had about 8 different addresses last time I looked and the IP it was using seemed to be different every day. My Linux server has a fixed IP as it's running a web server.

---

It's probably got MAC address randomisation turned on, and using a different MAC address, and the host part of the IPv6 address derived from it, every time it reconnects to the WiFi network. We are always having to turn this off on test devices so the test servers can find them.

[#14]

I never noticed that - I'll check next time the connection drops.
I have to keep editing my Think Broadband monitor when the drops happen.

[#15]

First of all, everything's been great so far on the IPv6 trial...

About the only feature I'm missing is reverse DNS.

High touch users get (or at least got?) reverse DNS for their IPv4, but there's nothing for IPv6 at the moment.  Perhaps something like this could work for high touch users?

Just thinking out loud, so any thoughts on this would be welcome...

The lowest 4 bits from their allocation (the first /60) gets a blanket reverse DNS which is the same as their IPv4 reverse DNS (username.plus.com), or otherwise some derivation of it - maybe (<an-ip-address-representation>.username.plus.com).  Lowest 4 bit because IPv6 reverse delegation happens on 4 bit boundaries.

The remainder of the reverse lookups in their IPv6 allocation gets delegated to a DNS server within their allocation itself (e.g. on the address with host part ::5353 within the highest numbered /64 of their prefix).  If there is no DNS server running on that address, then the reverse DNS lookup fails.

This would allow power users to run their own DNS without having to build any sort of UI or have any support interactions.

I'm not sure what other UK ISPs do, with the exception of Andrews and Arnold (who do everything of course :-).  https://support.aa.net.uk/Reverse_DNS

Might be a good question to ask at a UK IPv6 Council event...

Cheers,

Tim.