OpenSAND can be plugged behind an internet connection. Some configuration is required in order to be able to connect the internet from any terminal through OpenSAND.

1. Open /usr/share/opensand/topology.conf file and change the value sarp > default from -1 to 0.
   (this is the terminal ID for the GW, change this value if you want to plug a terminal as an access point to the internet).
2. Configure the default route from terminals toward the access point (Be careful if you are behind a SSH connection, you may lost your connection by replacing the default route. Please consider adding a route for this SSH connection).
   If you plugged the GW to the internet simply add this command
   1. To Satellite Terminal:

      # ip route add default via <GW lan address>

   2. To Work Station:

      # ip route add default via <ST/GW lan address>

3. Enable IP forwarding on GW. Set /proc/sys/net/ipv4/conf/all/forwarding value to 1
4. Add NAT on GW.
   

   # iptables -t nat -A POSTROUTING -o <GW interface plugged to the internet> -j MASQUERADE

5. If you are using a workstation:
   
   • Set the default route toward the associated terminal.
   • If your DNS server is local, check that you can still access it from the WS, else add a public one.

NB. you can avoid workstation configuration by enabling DHCP with default route when configuring the daemon on the satellite terminal.

The interface for satellite terminals or gateways LAN (default eth1) can be replaced by a virtual interface or by the opensand_tun interface itself.

However, you won't be able to plug workstations behind the terminals or gateways.

OpenSAND can be run without any special configuration on virtual machines. The platforms on which OpenSAND was tested and worked include: libvirt (with QEMU and KVM hypervisors), OpenStack, VMWare and VirtualBox. Concerning OpenStack, special attention must be made to network security, since OpenSAND configures IP addresses independently of OpenStack.

OpenSAND was tested and proved to run on Docker containers. However several aspects must be considered in order to make it work:

• the operation of OpenSAND (with services such as the OpenSAND daemon and OpenSAND collector) is incompatible with the philosophy of Docker, which favours the execution of foreground programs. To adapt OpenSAND to this philosophy, the OpenSAND daemon and collector must be run in foreground and not as services; for this to work, some scripts must be run beforehand in order to generate all necessary configuration files.
• OpenSAND needs multiple isolated networks between the different components: an emulation network linking all SAT, STs and GWs, and a LAN network for each ST and GW. Docker networks of type bridge must be created, with the proper subnet addresses, and carefully attached to the containers (in order for the configuration of interfaces of OpenSAND to work).