* Think how the DTLS part can use better negotiation of algorithms and DTLS
  versions than the current openssl string approach (using PSK ciphersuites
  seem to be like a solution, but then we could not use the session ID to
  forward the UDP connection to the proper worker).

* Try adding salsa20-12 and UMAC as encryption algorithms for DTLS to reduce 
  CPU load in systems without AES accelerator.

* Handle users being in multiple groups.

* Certificate authentication to the main process. Possibly that is just
  wishful thinking. To verify the TLS client certificate verify signature one
  needs instead of the signature, the contents of all the handshake messages, 
  and knowledge of the negotiated TLS version, in addition to being able to
  select the server hello random. That could be done sanely only if gnutls 
  provided facilities to set the server hello random, and override the client
  signature verification at an early stage before data are hashed (to verify
  that the set random value was present in the handshake).

* When a TUN device is in use and cannot be assigned mark it as such and
  continue.

* When a user (IP) gets into the BAN list multiple times, disable it for
  long.

* Change into a hashtables the lists that are used during a client
  connection.