Capwap Mailing List: nat, dtls and bind

[mauro . bis (mauro.bis]

Hello, I have a few questions about the 8th version of the CAPWAP draft.

1. Chapter 11 describes three possible scenarios involving NAT.
 My question is about the second scenario, where more than one WTPs is
 deployed behind the same NAT system. The text says that thanks to the
 "Session ID" message element of the Data Channel Keep Alive Packet the
 AC can distinguish the control and data flows of NATted WTPs, but how
 can the WTPs receive possible responses to their requests since the
 AC will address any response to the same IP address (the NAT box IP)?
 Do they need to employ UDP "hole punching" techniques?

2. In chapter 2.4 there is the statement "...there are DTLS library
 implementations currently available..." but I can only find the
 OpenSSL one. May someone please point me to other (possibly free)
 implementations?

3. Section 4.4.1 says that the Data Channel Keep Alive packet is used
 to bind the control channel with the data channel.
 However it is not clear to me the way in which the Keep Alive packets,
 sent on the data channel, relate to the control channel.
 What is the meaning of the term *bind* in this context?

Thanks in advance,
Mauro