IPv6 Transition Technology Survey

Dual stack is going to be a requirement for the foreseeable future because a large number of people don't even have IPv6 connectivity.  I use Comcast in the United States, and I'm a Comcast business customer.  I have written plenty on the forums.businesshelp.comcast.com forums on how to get IPv6 working properly on their default business modem the Technicolor.  There is, now, a significant community of people using IPv6 - and that's STATICALLY ASSIGNED IPv6, on that network - and yet, Comcast STILL has problems - for example they released a buggy firmware update for those modems at the beginning of February and it took weeks of calling by many angry customers who's servers were no longer accessible via IPv6 before Comcast acknowledged the issue and released a patch (the patched and working firmware is CGA4332COM_6.7p9s1_PROD_sey if anyone cares)  All of this is because of the IPv4-centric viewpoint of high tech in the United States, which has the lion's share of IPv4, which in Comcast's case didn't even include a complete IPv6 testbed on firmware releases on their primary business class cablemodem.

I'm glad at least that Microsoft has someone who cares enough to even ask about this - although it's not hardly like this survey was even published very much.

It also would be an excellent move to STOP emphasizing DHCPv6 in these discussions because that was NEVER the primary method of address assignment envisioned by the designers of IPv6.  SLAAC is the intended method of IP address assignment and RFC 8106 has defined assignment of DNS servers for Internet lookups and Multicast DNS for local LAN name lookups.  SLAAC and RFC8106 and mDNS are all supported by Windows 11.  The only reason people still cling to the name DHCP is because it came from IPv4 and for many years the dual-stack paradigm was emphasized with no support for automatic assignment of IPv6 DNS server addresses, the expectation was that people would ALL dual-stack, use DHCPv4 for IPv4 assignments, SLAAC for IPv6, and IPv4 DNS servers would be mandatory.  DHCPv6 was developed by people wanting to jump the gun and it's primary boost was because it used the same acronym as DHCPv4.  but now that we have a standard DNS RA, the use of DHCPv6 should be de-emphasized.

> The journey toward IPv6 only networking is challenging, and today there are several different approaches to the transition from IPv4 to IPv6 with multiple dual-stack or tunneling stages along the way.

How much of the challenge is due to all the IPv4-to-6-back-to-4-then-maybe-6-again transformation you have to do to make them work concurrently and interchangeably? Because my understanding is that if you simply ditch IPv4 wherever you can, it would tremendously simplify a lot! No tunnelling, no NAT, no dual stack, etc. etc. etc.

Hopefully, you'll make your survey statistics public so our business can soon learn what's important.

Hello tojens 

Thanks for putting this Technology Survey forward.

Some important points to mention about this topic for Windows to improve are the following:

- CLAT support to any Network Interface and not only for WWAN

- Support for RFC8781 (Discovering PREF64 in Router Advertisements)

- Support for RFC8925 (IPv6-Only Preferred Option for DHCPv4)

- Complete support for PCP (Port Control Protocol) to automate the creation of IPv6 Pinholes for applications that require incoming connections (similar to what was previously done by UPnP for IPv4)

As an extra thing that is not necessarily related to IPv6 Transition but it is certainly a core network thing to fix is support for /31 prefixes (RFC3021), not only on Windows 11 but on the previous still supported versions.

Regards