If the communication link between the database servers fails, both database servers might assume that the other server is down. This can lead to both servers acting as the primary server, and you might lose transactions when databases are later synchronized. To avoid this situation, you can use a network reference device, an External Reference Entity (ERE) to check the health of the network.
If an ERE is used then, when the HAC cannot detect the remote server, the HAC checks the status of the physical link between the local HSB server and the ERE device by pinging the ERE. If the physical link to the ERE is not operational, the HAC sets the local server to SECONDARY ALONE state. If the physical link is operational, the HAC sets the local server to PRIMARY ALONE state. Consequently, an HSB server that cannot connect to the ERE becomes the secondary server, and the situation where both servers become the primary server is avoided.
▪ any computer in the network outside the cluster.
If a redundant network (that is, duplicate network controllers, cables and switches) is used in a cluster, define ERE outside the cluster.
Note If the HSB link is considered unreliable (including all the cases where ERE is used), the HSB server parameter HotStandby.AutoPrimaryAlone should be set to no.
ERE must use the same HSB link that is used by the keepalive messages.
