Deploy for disaster recovery
The deployment topology to provide availability in the event of an availability zone failure is similar to the high availability topology. An active/active pattern is still used for Liberty, Solr, and ZooKeeper, however the servers that you deploy are in separate availability zones.
The components behave in the same way as in the high availability deployment topology, except they are deployed in availability zones. You can think of an availability zone as a data center. By deploying across a number of availability zones, the deployment continues to function if one of the zones fails.
Note: For Microsoft SQL Server, you must have a third database server to act as a witness. To continue to provide availability when one of the zones fails, this server must be located outside of the two availability zones. For more information, see Three synchronous replicas.
- Because ZooKeeper requires more than 50% of the servers in an ensemble to be available, one server must be located outside of the availability zones. The ZooKeeper server does not require a significant amount of resources so it can be located on any server that can communicate with the availability zones.
- You can deploy multiple Liberty and Solr servers to provide more availability at a server level within each zone.
- For the components that use an active/active pattern, servers in each availability zone are used to process requests. To maintain the performance of the system, ensure that the network that is used between availability zones is stable and provides enough bandwidth for system operation.