One Identity Safeguard for Privileged Sessions 6.0.1 - Deployment in Single-Interface Router Mode

Advanced or Policy-based routing

Usually there is a central network device somewhere close to the location where SPS is planned to be implemented. This central network device (a router or firewall) can facilitate improving the previous layout into a real working scenario, as it is visible on Figure SPS in single-interface transparent mode.

There is a router (or firewall, or layer3 switch) between the zones, and SPS is installed into a new, separated network. Here, all the devices (including SPS) are configured to use the central router as their default gateway. So, if a client is trying to reach a server, the connection is going through the router. However, SPS is not able to audit the remote administration with this configuration.

Here comes the router into the play. For example, if we have to audit RDP connections, the router can be configured to route all the connections to SPS (connections are coming from the client network, going to the server network, and the destination port is 3389). With this configuration, all RDP connections are “redirected” to SPS. SPS sends the traffic back to the router, which then sends the connection to its original destination, to the server zone.

SPS receives a connection on its “External” interface, and it routes it back on the same interface. It creates a “hook” in the network traffic, but it also makes the connection totally transparent: the client IP can be the same (optional), and the client does not have to know anything about SPS. Non-administrative traffic can also be unaffected, as we can selectively route the necessary traffic to SPS.

The configuration of the router can vary depending on the type of the router. The following two procedures describe configuration scenarios for a Linux and a Cisco router.

Example scenarios

On the SPS side, no special configuration is required. SPS has to be in transparent mode. The “External” interface has to be configured correctly, and has to be connected to the router. On the “Internal” interface any IP address can be configured that is not used on the network, as we will not use this interface at all. The default gateway should be the router.

Configuring advanced routing on Linux

The following describes how to configure a Linux-based router to redirect selected traffic to One Identity Safeguard for Privileged Sessions (SPS) instead of its original destination. This procedure should work on most modern Linux-based routers, including Check Point firewalls.

To configure a Linux-based router to redirect selected traffic to SPS instead of its original destination

1. Create the packet filter rules that will mark the connections to be sent to SPS using the CONNMARK feature of iptables. Mark only those connections that must be redirected to SPS.

   # iptables -t mangle -I PREROUTING -i <interface-facing-the-clients> -p tcp -d <network-of-the-servers> --dport <port-to-access> -j CONNMARK --set-mark 1

   Example: Setting up a connection mark for Linux policy routing

   For example, if the network interface of the router that faces the clients is called eth0, the servers are located in the 10.0.0.0/24 subnet, and the clients access the servers using port 3389 (the default port of the RDP protocol), then this command looks like:

   # iptables -t mangle -I PREROUTING -i eth0 -p tcp -d 10.0.0.0/24 --dport 3389 -j CONNMARK --set-mark 1

2. Create a rule that redirects the answers of the servers to SPS. That way both the client-to-server and the server-to-client traffic is routed to SPS.

   NOTE: This step is only required if you want to use Source NAT (IP Spoofing) instead of SPS’s address towards the monitored servers.

   Figure 3: Control > Connections — Using SNAT

   

   # iptables -t mangle -I PREROUTING -i <interface-facing-the-servers> -p tcp -s <network-of-the-servers> --sport <port-to-access> -j CONNMARK --set-mark 1

3. Convert the CONNMARK marks to MARK:

   # iptables -t mangle -A PREROUTING ! -i <interface-facing-the-scb> -m connmark --mark 1 -j MARK --set-mark 1 

   Caution: This rule must be placed after the CONNMARK rules.

4. Add the table name to the /etc/iproute2/rt_tables of the router. Use the following format (for details on routing tables, see for example the Guide to IP Layer Network Administration with Linux):

   103 scb

5. Create a routing table that has a single entry with a default route to SPS:

   # /sbin/ip route add default via <ip-address-of-SPS> table scb

6. Create a routing rule that selects the routing table called scb, if the connection is marked.

   # /sbin/ip rule add from all fwmark 1 table scb

7. If SPS is configured to spoof the IP address of the clients on the server side (that is, the SNAT > Use original IP address of the client option of the connection policies is selected), enable spoofing on the router for the interface connected to SPS.

   # echo 0 > /proc/sys/net/ipv4/conf/<interface-facing-SPS>/rp_filter
   # echo 0 > /proc/sys/net/ipv4/conf/all/rp_filter

   Expected result:

   The traffic from the clients targeting the specified port of the servers is redirected to SPS. Therefore, SPS can be configured to control and audit this traffic.

Configuring advanced routing on Cisco routers

The following describes how to configure a Cisco router to redirect selected traffic to One Identity Safeguard for Privileged Sessions (SPS) instead of its original destination. This procedure should work on most modern Cisco IOS releases but was specifically tested on IOS version 12.3.

To configure a Cisco router to redirect selected traffic to SPS instead of its original destination

1. Create an ACL (Access Control List) entry that matches the client and server subnets and the to-be-audited port. Keep in mind that whatever is permitted by this ACL is what will be matched, so make sure that the scope of the ACL entry is narrowed down as much as possible.

   #(config) ip access-list extended ssh-inbound
   #(config-ext-nacl) permit tcp <src net> <src mask> <dst net> <dst mask> eq <dst port>

   Example: Configuring an ACL entry for Cisco policy routing

   For example, if the clients are in the 192.168.0.0/24 subnet, the servers are located in the 10.0.0.0/24 subnet, and the clients access the servers using port 22 (the default port of the SSH protocol), then the permit clause should be:

   #(config-ext-nacl) permit tcp 192.168.0.0 0.0.0.255 10.0.0.0 0.0.0.255 eq 22

   TIP: Cisco ACLs use inverse netmasks for defining network addresses. To calculate an inverse mask given a subnet mask, simply subtract each octet value from 255.

2. Create an ACL entry that matches the reply packets coming from the server zone and targeted at the client zone to make sure that replies are reaching the SPS.

   #(config) ip access-list extended ssh-outbound
   #(config-ext-nacl) permit tcp <dst net> <dst mask> eq <dst port> <src net> <src mask>

   NOTE: This step is only required if you want to use Source NAT (IP Spoofing) instead of SPS’s address towards the monitored servers.

   Figure 4: Control > Connections — Using SNAT

   

   Example: Configuring an ACL entry for reply packets with Cisco policy routing

   In case of the example in step 1, the permit clause should be:

   #(config-ext-nacl) permit tcp 10.0.0.0 0.0.0.255 eq 22 192.168.0.0 0.0.0.255

3. Create a route-map entry. It controls which packets are affected by policy routing and where they should be forwarded to. The match commands specify the conditions under which policy routing occurs. The set commands specify the routing actions to perform if the criteria enforced by the match commands are met. A new route-map can be defined as follows:

   #(config) route-map scb-inbound

   1. Set your route-map to match the traffic in ACL ssh-inbound:

      #(config-route-map) match ip address ssh-inbound

   2. Set an action on the matching traffic. Define a next-hop entry to redirect the traffic to the SPS.

      #(config-route-map) set ip next-hop <SPS IP address>

4. Create another route-map that controls the reply packet flow.

   #(config) route-map scb-outbound
   #(config-route-map) match ip address ssh-outbound
   #(config-route-map) set ip next-hop <SPS IP address>

   NOTE: This step is only required if you want to use Source NAT (IP Spoofing) instead of SPS’s address towards the monitored servers.

   Figure 5: Control > Connections — Using SNAT

   

5. Apply the route-map to the appropriate interfaces.

   1. First, add the ssh-inbound route-map entry to the interface facing the clients:

      #(config) interface <interface-facing-the-clients>
      #(config-if) ip policy route-map scb-inbound

   2. Then add the ssh-outbound route-map entry to the interface facing the servers:

      #(config) interface <interface-facing-the-servers>
      #(config-if) ip policy route-map scb-outbound

   Expected result:

   The traffic from the clients targeting the specified port of the servers is redirected to SPS. Therefore, SPS can be configured to control and audit this traffic.

   The full configuration for the above topology:

   ! interface facing the clients
   interface FastEthernet0/0
    ip address 192.168.0.254 255.255.255.0
    ip policy route-map scb-inbound
    duplex full
    speed auto
    no mop enabled
   
   ! interface facing the SCB
   interface FastEthernet0/1
    ip address 172.16.0.254 255.255.255.0
    duplex full
    speed auto
    no mop enabled
   
   ! interface facing the servers
   interface FastEthernet1/0
    ip address 10.0.0.254 255.255.255.0
    ip policy route-map scb-outbound
    duplex full
    speed auto
    no mop enabled
   
   ! access lists matching the server and client subnets and the SSH port - incoming packets
   ip access-list extended ssh-inbound
    permit tcp 192.168.0.0 0.0.0.255 10.0.0.0 0.0.0.255 eq 22
   ! access lists matching the server and client subnets and the SSH port - reply packets
   ip access-list extended ssh-outbound
    permit tcp 10.0.0.0 0.0.0.255 eq 22 192.168.0.0 0.0.0.255
   
   ! policy routing entry matching on the incoming SSH connections and
   ! redirecting them to the SCB external interface
   route-map scb-inbound permit 10
    match ip address ssh-inbound
    set ip next-hop 172.16.0.1
   
   ! the following part is only required for SNAT-based SCB configuration
   ! policy routing entry matching on the SSH reply packets and
   ! redirecting them to the SCB external interface
   route-map scb-outbound permit 10
    match ip address ssh-outbound
    set ip next-hop 172.16.0.1