Dynamic tunneling is a transparent mechanism available for applications that support the SOCKS4 or SOCKS5 client protocol. Instead of configuring port forwarding from specific ports on the local host to specific ports on the remote server, you can specify a SOCKS server which can be used by the user's applications. Each application is configured in the regular way except that it is configured to use a SOCKS server on a localhost port. The Secure Shell client application opens a port in the localhost and mimics a SOCKS4 and SOCKS5 server for any SOCKS client application.

When the applications connect to services such as IMAP4, POP3, SMTP, HTTP, and FTP, they provide the necessary information to the SOCKS server, which is actually the Secure Shell client mimicking a SOCKS server. The client will use this information in creating port forwarding to the Secure Shell server and relaying the traffic back and forth securely, as with user-specified port forwarding.

With sshg3 on the command line, the syntax of the dynamic tunneling command is the following:

client$ sshg3 -L socks/[listen-address:]listen-port server

For example, the following command will set up dynamic tunneling from port 1234 on the client to sshserver. The applications are set to use a SOCKS server at port 1234 on the client. From the server, the connections are forwarded unsecured to the destination hosts requested by the applications.

sshclient$ sshg3 -L socks/1234 username@sshserver

Dynamic tunnels can also be defined for connection profiles in the Connection Broker configuration file. The following is an example from a ssh-broker-config.xml file:

<profile id="id1" host="sshserver.example.com">
...
  <tunnels>
    <local-tunnel type="socks"
                  listen-port="1234"
                  allow-relay="no" />
  ... 
  </tunnels>
</profile>