5.3.2 Considerations on using multiple servers
You must consider the points discussed in this subsection when you use multiple servers in a hot-standby configuration.
- Organization of this subsection
(1) Whether multiple servers are used to perform a single job
If multiple servers are used in tandem to perform a single job, evaluate the use of the grouped-system switchover function that groups multiple servers together and performs hot standby processing.
When multiple servers are grouped together, they are treated as a group of servers. If a failure occurs on one of the servers in the group, HA Monitor performs hot standby processing for the entire group. You can avoid unneeded hot standby processing because you can base whether hot standby processing is to be performed on the importance of each server.
For details about grouping servers together and the required environment settings, see 3.2.3 Grouped-system switchover achieved by grouping servers.
(2) Whether it is necessary to determine the server start order
If you need to determine the start order of multiple servers, evaluate the use of the server switchover order control function.
You can group multiple servers and determine the order in which the servers in the group are to be started and terminated during hot standby processing in the event of a failure. For example, if a group of servers to be placed includes a required server, you can specify settings in such a manner that the required server is started first and then the remaining servers are started.
The server switchover order control function does not apply to the order in which servers are started during server startup. Therefore, during server startup, the operator must manually start the servers in the predefined start order. If you want to automatically control the start order during server startup, evaluate running servers in the monitor mode and treating multiple programs as a single server.
For details about server switchover order control and the required environment settings, see 3.2.4 Controlling the server switchover order.
(3) Whether a specific host has too much load after hot standby processing
In a configuration in which a secondary system is shared, such as the 2-to-1 switchover configuration, a large amount of job processing might be concentrated on a single host after hot standby processing. HA Monitor provides a lock function that terminates the standby servers in a secondary system whenever the active server in the secondary system is started. If you want to improve system performance by preventing a heavy load from being placed on a specific host, consider using this lock control function. This function enables you to prevent a concentration of load because it can limit the servers to be run on a single host.
We recommend that you use this lock function in the n-to-1 switchover configuration (n: 2 or greater) in which loads tend to be concentrated in the secondary system. For details about the lock function, see 3.2.6 Locking servers to prevent concentration of the processing load after the hot standby operation.