Why bother, when I could just do TLS SNI reverse proxying via nginx?
* Some services don't use TLS, or even TCP.
* A reverse proxy is yet another intermediary in the chain.
* Plain IPv6 routing is simpler than reverse proxying, and I already need a network layer anyway.
There are downsides:
* some software doesn't support IPv6. I haven't experienced this on the Linux servers I run.
* in a dual stack network, now you have two networks! I use NAT64/PREF64 like https://labs.ripe.net/author/ondrej_caletka_1/deploying-ipv6... to have most clients only be on IPv6. They get IPv4 connectivity over IPv6 via NAT64.
* If I'm in another country then I often don't have IPv6 connectivity. In this case I use any VPN that offers IPv6 (and have one available via my home, via Wireguard).
* Learning IPv6 takes time, but not much. It's one-off. It's not more complex than IPv4, but it is different. If anything, it's simpler. (SLAAC rather than DHCPv4; IP reachability rather than NAT/port forwarding).