In large networks the performance of SysOrb can be improved by clustering several SysOrb Servers. The most important reasons for using ServerCluster are:
Performance: Since the servers in a ServerCluster distribute the needed Agent checkins and NetChecks between them, checks can be performed at a higher rate.
A ServerCluster provides full mirroring of all data in the database - one should therefore be aware, that performance problems directly related to the I/O subsystem of one SysOrb Server, cannot be solved simply by moving to a ServerCluster setup.
Redundant servers: If one of the SysOrb Servers in the cluster somehow fails or disappears from the network (e.g. due to hardware faults) the other SysOrb Servers notice this and fill in for the missing server. On the agent side, SysOrb Agents will cycle through their list of SysOrb Servers if they are unable to contact a server.
If the connection between two groups of SysOrb Servers in the cluster is lost, each group will perform all checks on the clusters NetChecks and snmpChecks. When the connection is restored, the two groups will synchronize their databases using the best results for the tests.
To setup a number of SysOrb Servers as a cluster, each SysOrb Server needs to have a list of all the SysOrb Servers in the cluster, including itself. The hosts must be mentioned in the same order, on all SysOrb Servers, or the cluster will not work correctly. The cluster members can run different operating systems, but must be running exactly the same version (including build number) of the SysOrb Server software.
See the Server Configuration section for more details on how to configure the SysOrb Servers.
Note: Clustered SysOrb Servers do not automatically monitor each other, but this can quite easily be achieved by setting up the SysOrb Servers as SysOrb Agents as well. It is highly recommended to always run a SysOrb Agent on the SysOrb Server machine.