Overview of NIC Host Redundancy
In NIC host redundancy, two NIC hosts with the same configuration act as a redundant pair. One host assumes the active role, and the other the passive role. If the active host fails, the passive host writes its Common Object Request Broker Architecture (CORBA) object reference to the directory and assumes the active role. The former active host reassumes the active role only when the current active host fails. Each pair of redundant hosts constitutes a community.
You can also add a monitor, which tracks the redundant components, to that community. The SDX software includes a NIC monitor component; you can install one NIC monitor on a machine. You can then add that monitor to the directory and use it in a host redundancy configuration.
The monitor acts on behalf of a client or server that is not part of the community. If the client or server cannot communicate with the active host in the community but the hosts can communicate with each other, the monitor prompts the passive host to assume control. Without the monitor, the passive host would not assume control, and the host would be unavailable.
For example, in the network configuration for the OnoPopAllRealms distributed configuration shown in Chapter 13, NIC Configuration Scenarios, the monitor BOClient is installed on the same machine as the NIC proxy and tracks the communications between the hosts OnePopBO/One and OnePopBO/Two. If the monitor detects that the connection to the active host is unavailable, it prompts the passive host to assume control.
The members in the community exchange keepalive messages to monitor each other's availability. If a member does not receive a keepalive message from another member during the specified time, the first member assumes that the second member is unavailable.