Being a member of the regional NIC isn't that big a deal. Much of the time we've only been using provider blocks because the NIC's policy won't let us get provider-independent space without significant justification, or because the provider just won't permit provider-independent blocks to be advertised through them. The former goes away with IPv6, and the latter can be solved by switching to a provider that is more sensible about customer requirements.
Redundant Internet connections are going to need portable space, just like they did in IPv4 before the NAT hack was added to the protocol. Per my previous paragraph, getting that isn't going to be tough. This is the IAB's recommended approach, per the RFC 5092.
Without NAT, local applications can read the IP address directly from the NIC, should they need it. Remote applications can just use DNS.
As for DNS records, DHCPv6 in combination with IPv6 auto-configuration handles DNS updates dynamically.
I have yet to see a need for port remapping where more than a single global IP address is in play. The smallest block allocated under IPv6 is a /64, which means that you can assign an IPv6 address for every service that you want to make available and then move the global IP of that service from machine to machine as needed. It's a different paradigm.
IPv6 has been around in test for a decade. NAT was never needed in IPv4 until people started worrying about IP address depletion. It will likely never be needed with IPv6.