Address Manager v9.0.0 introduces enhancements to database replication through streaming replication.
In previous releases, database replication used file-based log shipping of WAL files to propagate information from the Primary server to the Standby Servers. Configuring database replication would require you to configure a disk usage-based WAL file threshold. When WAL file threshold limits were reached due to the inability for the Standby server to process the WAL files, the Primary server would break replication with the Standby servers. Performing a failover would cause the Primary Address Manager server to reboot, resulting in service outages. If an update was required to the Address Manager servers in replication, you would need to break replication, apply the update to each Address Manager server, and recreate the replication environment.
With streaming replication, customers benefit from near-zero-downtime during failover and more up-to-date information on the Standby servers, resulting in a quicker failover and increased reliability during disaster recovery scenarios. The disk usage WAL file threshold has been removed and replaced with latency thresholds, which is defined by how far a Standby lags behind the Primary servers. If latency thresholds have been reached for a specific Standby server, an alert is created through the event log or SNMP trap, notifying the administrator that there's an issue with the replication set-up and allowing them to take actions as they see fit. When a failover is performed, the Primary Address Manager server doesn't reboot and services remain available, including API services. Streaming replication also introduces a graceful failover, allowing the replication latency to reach 0 seconds before the failover is performed. If the Address Manager administrator needs to perform the failover immediately, the graceful failover can be overridden and the administrator can perform a forced failover. Streaming replication also introduces the ability to apply Address Manager updates to all servers in replication without the requirement to break replication. The update can be applied to the Primary server and the update will also be applied to the connected Standby servers.
For more information on database replication, refer to Replicating the database for Address Manager disaster recovery.