Skip to content

Plan your network

Take five minutes before you install. The installer asks one network question first (the management interface), and then one interface per decoy.

The topology rule

One decoy per interface, on that interface's own IP

  • Each decoy gets its own network interface and answers on that interface's own IP address and MAC address, like a real server.
  • An interface holds at most one decoy. An interface used by one decoy is not offered for the next.
  • The manager may share an interface with one decoy. On a server with one network card, that is the only option.

So a server with one network card holds one local decoy; a server with three cards (one for management) can hold three. More decoys in other networks come from remote sensors, within the limit of 3 decoys in total.

Enterprise feature: available in ReD-Pot Enterprise

Flexible decoy placement and more decoys. Enterprise can also place a decoy on an extra free IP address of a shared interface (on its own link with its own MAC), in a tagged VLAN, and with a VMware, Hyper-V, Dell, Xen or QEMU MAC prefix. Several decoys can then share one network card without moving host SSH. The number of decoys is set by the licence instead of the Community limit of 3.

About ReD-Pot Enterprise · Compare editions · Contact us for a licence

Three questions

  1. Which interface do you use to manage the server? (your SSH session). That becomes the management interface. The web UI, the sensor API and host SSH answer only there.
  2. Which networks should the decoys appear in? Usually your server and user VLANs, where an intruder would land. Each decoy needs its own interface in that network.
  3. Which servers should they pretend to be, with which names? Up to 3 decoys in total (local decoys plus remote sensors). Use your real naming scheme (RS-FS-02, not FS01). Host names are 1 to 15 characters (Windows limit).

Typical setups

Your situation Choose
Office or data centre, server with 2 to 4 NICs management on one NIC, one decoy on each other NIC (recommended: attackers on the decoy networks never see the admin ports)
Server with one NIC inside the LAN one decoy on that NIC; it shares the management IP and host SSH moves to a random port. Add more decoys as remote sensors
VPS with one public IP (internet-facing) the same as one NIC: one decoy on the server's address, host SSH moves
Branch office or another VLAN install a remote sensor there that reports to your manager

What happens to host SSH

Interface you pick for a decoy What it looks like Host SSH
Another interface (not management), recommended that NIC is the decoy; attackers on that network never see the admin ports stays on port 22 and listens only on the management IP
The management interface (the only choice on a 1-NIC server) the decoy shares the IP with your SSH and the web UI moves to a random port, because OpenSSH on 22 would give away a "Windows" decoy
 One interface: management + one decoy            Two interfaces: management on one, decoy on the other
 +---------------- server ----------------+     +---------------- server -----------------+
 | eth0 10.0.0.5 (management + decoy)     |     | eth0 10.0.0.5        | eth1 10.0.50.10  |
 |  +- FS01 :135 :139 :445 :3389          |     |  +- SSH :22          |  +- FS01 :135    |
 |  +- web UI :5xxxx, sensor API :5xxxx   |     |  +- web UI :5xxxx    |     :139 :445    |
 |  |   (allowed sources only)            |     |  +- sensor API       |     :3389        |
 |  +- SSH moved to a random port         |     |   management only    |  (eth1's own IP) |
 +----------------------------------------+     +----------------------+------------------+

On a decoy's interface only the decoy answers

The host firewall accepts the decoy's own ports on its interface and answers everything else the way a closed port does (TCP reset, ICMP port unreachable, on IPv6 too). This hides every other service on the server from that interface, including the management interface when a decoy shares it: a monitoring agent or another web server listening there is no longer reachable through it.

When a decoy shares the management interface:

  • the web UI, sensor API and host SSH allow-lists default to your own SSH client address (/32);
  • the installer lists the host services that listen on that interface and become unreachable there, and asks you to confirm.

See Ports and firewall and Access allow-lists.

Addresses

Give the management interface, every decoy interface and every remote sensor a static address or a DHCP reservation. The manager's sensor-API allow-list does not follow a new DHCP lease of a remote sensor; the installer warns when an address it binds to is a DHCP lease.