Failover of the Data Service PICs
The data PIC failover procedure for the BGF assures service continuity in case of a service PIC failure. The BGF architecture provides both a 1:1 redundancy model and a 1:N redundancy model for data service PICs.
In the 1:1 redundancy model, there is one primary service PIC and one secondary service PIC that acts as a backup. If the primary service PIC fails, the Routing Engine allocates the secondary service PIC, and the software switches over to the secondary service PIC. The pgcpd process continues to provide the PGCP services as if the service PIC has restarted. All states and sessions are lost, but new calls are accepted. When the failed service PIC recovers, it does not take over from the redundant PIC.
In the 1:N redundancy model, you can define one PIC as the secondary of many primary PICs. In this model, after the secondary PIC becomes active, all other primary PICs are left without a secondary PIC. Even when the primary PIC recovers, it does not become a redundant PIC to all of the primary PICs. A recovered primary PIC can replace only the same secondary PIC that previously replaced it. This functionality means that administrative involvement is usually required after a failover event happens.
Procedure in Case of Data PIC Failure
If a data PIC fails, the following series of events occurs:
- Active calls are lost.
- The BGF notifies the gateway controller of the failure using an FO/904 ServiceChange message.
- The BGF receives an acknowledgment for the FO/904 message from the gateway controller.
- The redundant service PIC (rsp) mechanism allocates the secondary PIC and makes the secondary PIC the new primary PIC.
- The new primary PIC establishes the IPC connection to the pgcpd process on the Routing Engine.
- The pgcpd process issues an RS/902 registration message to the gateway controller.
- The pgcpd process receives an acknowledgment of the registration message from the gateway controller.
- The BGF is ready to accept and process new H.248 commands.
