[gumby]

FTP needs no defending -- it was really useful in 1979, but times have changed (e.g. I suspect every machine on the Internet uses an 8-bit byte). One point you wrote surprised me though:

> no clear client/server relationship which can cause issues for NATing and filewalls (particularly if running with TLS)

Really, crocks like NAT and stateful firewalls should die. Layers 4 and below are inherently peer-peer -- the net should not treat endpoints differently (i.e. should not privilege some over others). That simply encourages a "client" or "consumer" mentality in both the technical and social senses.

[martinald]

The thing is while NAT is horrible for what you're saying, it probably did more to improve security than anything else, which wasn't it's primary goal.

I remember what the internet was like when ADSL/cable models first came along. Everyone was getting pwned none stop. Any RCE could easily be applied by scanning a consumers DSL/cable IP pool and you'd be able to hit a very high %age of them.

NAT totally stopped this.

[laumars]

It was the firewalling that stopped those attacks. Granted you could argue that the firewalls only came popular in households because routers were shipped to address a need for NATing but pragmatically we really should have been installing firewalls on our PCs in the pre-router days of the internet.

[gumby]

> The thing is while NAT is horrible for what you're saying, it probably did more to improve security than anything else, which wasn't it's primary goal.

Are you defending NAT? It sounds like a Vietnam era construction: you had to destroy the Internet in order to save it.

We now have a seemingly entrenched tree-structured (i.e. centralized) network again, the very 1960s architecture we tried so hard to get away from.