VS>> The security guidelines I have read don't specify "NAT must be VS>> used." They specify "RFC1918 addresses must be used in the VS>> internal network."
MR> For IPv6 they could use ULA (RFC4193). ;)
Good point. Thank you. Maybe fc00::/7 has a chance of becoming the new 192.168/16.
VS>> A static NAT has limited usage and indeed does not provide much VS>> additional security. But the dynamic NAT and especially PAT VS>> provide a very important security feature no packet filter VS>> provides: they *hide* the *source* *addresses* of internal hosts VS>> thus effectively hiding the network structure from outsiders.
MR> And some dumbass enables UPnP on the firewall/router. >:)
I don't think enterprise-class firewalls have UPnP, do they?
And thinking about SOHO and home routers/firewalls, what kind of IPv6 connectivity are they going to have, what do you think? Those present who have native IPv6 connectivity, what's your ISP's policy on assigning addresses to customers?
If my ISP were going to give me one IPv6 address (a /128) or even one /64 net, this would be too few for my purposes. For my current home network, I use five /64s, so for me it would be a /56 at least.
MR> If an MR> organization thinks that it has to hide the internal IP addresses for MR> security reasons it can use NAT or proxies. Anyway, they still need MR> much more than that to secure their network.
MR>>> There's also NAT for IPv6.
VS>> Never heard of that, other than DNS64/NAT64 which are for a VS>> different purpose.
MR> NAT66
Interesting. Do you know of any implementations that could translate ULA addresses into one global /64 pool?
