That's a good list of hacks I've come across. But I thought we stopped using the 'symmetric nats' terminology a decade ago when the IETF Behave specification c

That's a good list of hacks I've come across. But I thought we stopped using the 'symmetric nats' terminology a decade ago when the IETF Behave specification came out? https://tools.ietf.org/html/draft-ietf-behave-rfc3489bis-18 https://tools.ietf.org/html/draft-ietf-behave-rfc3489bis-18 The specification tries to classify Nat behaviour into 3 classes: (1) when a port mapping can be reused (for any outgoing connection, for an outgoing connection to the same IP, for an outgoing connection to the same IP:Port endpoint), (2) what port is allocated when a new mapping is created (same as the private IP if possible, a contiguous port generated by a global counter, or a random port), and (3) how are incoming packets filtered/dropped based on the source IP:port (not filtered, ok from same IP, must be from same IP:port). Most Nats can be described using some combination of these behaviours. See this paper for details: https://www.hivestreaming.com/natcracker-nat-combinations-matter/ https://www.hivestreaming.com/natcracker-nat-combinations-ma...