One interesting thing to consider is the snat setting for the virtual server. Lets say that you
have two virtual servers inbound_vs and outbound_vs. Each virtual server has a route which uses a
corresponding transport config, inbound_tc and outbound_tc. Calls received by the inbound_vs
would be routed to connection created using the settings of inbound_tc. The persistence entry for
these calls would contain the inbound_vs as the source transport and the inbound_tc as the
Likewise calls received by the outbound_vs would be routed to connections created using the
setting of outgound_tc. The persistence entry for these calls would contain the outbound_vs as
the source transport and outboind_tc as the destination transport. If a call arrives on a
connection created via outbound_tc and a valid persistence entry still exists, it would route to
a connection using the outbound_vs transport. If no connection is found, it would create a new
outbound connection using the outbound_vs’s parameters.
Therefore, the virtual server SNAT setting should be that of the VLAN it is on. This is
opposite from traditional BIG-IP virtual servers.
Lets say that the inbound_vs listens on the external VLAN and the outbound_tc is for creating
connections on the external vlan. The inbound_vs’s SNAT settings are what would be used for
creating outgoing connections also on the external VLAN. Inbound_vs’s snat setting would never be
used for creating connection on the internal VLAN.
In this case, the SNAT settings of the inbound_vs should match the SNAT settings of the
outbound_tc. Likewise the SNAT settings of the outbound_vs should match the SNAT settings of the