A HotStandby configuration allows for a Primary server, Secondary server, and the Watchdog to reside in different machines and use different operating systems and APIs as shown in the example in System design issues. For details on implementing heterogeneous configurations, read System design issues.

All communication between the Primary and Secondary database (including putting a failed system back in service and re-synchronizing Primary and Secondary databases) occurs within existing communication layers, such as TCP/IP. HotStandby requires no auxiliary storage or transfer methods, such as shared disks or FTP transfers.

Important: If you are running the Watchdog on the same machine where the Secondary server resides, be sure to set the parameter AutoPrimaryAlone to no. In this situation, setting AutoPrimaryAlone to no is crucial because it prevents the potential error of having two primary servers. Primary may be in the PRIMARY ALONE state, and the Watchdog at server failure could switch Secondary to a PRIMARY ALONE state. This same error can also occur if a user happens to set the old Secondary server to become the new Primary. For more information about dual primaries, see Network partitions and dual primaries.