Virtual switching is supported at the application group level within instances licensed for Assure MIMIX HA. The instance cannot be configured as an INTRA environment.
Configuration requirements: The following configuration requirements must be met:
The application group must be of type non-clustered.
Target journal inspection must be configured and active on the node used for testing. Target journal inspection must be active for the system journal on the test node and for all user journals on the test node that are currently the target node of data groups within the application group.
All data groups associated with the application group must be switchable, must allow journaling on their target node, and must allow cooperative processing through the user journal.
Keyed replication is not supported and cannot be configured in any of the data groups or selection rules for libraries and files (object entries and file entries) associated with the application group.
Objects replicated by the data groups associated with the application group must be journaled on the target node.
You can use the procedure PRECHKVRT to check for configuration requirements in advance. The virtual switch procedure SWTVRT also checks configuration requirements and will not continue if a requirement is not met.
Operational considerations: There is exposure to your environment for the duration of a virtual switch and differences in some MIMIX behavior. Be aware of the following:
During a virtual switch, replication activity for your production environment is not complete because apply processes are stopped. Journal entries for replicated transactions accumulate in queues for apply processes on the target node. When the recovery phase of the virtual switch begins and the apply processes are started, their backlog may be at or exceed threshold conditions. It may take a significant amount of time to process a large backlog of transactions. The queued backlogs and retained IBM journals for the amount of replication activity and virtual switch activity that occurs during a virtual switch may cause DASD issues.
For the duration of the virtual switch procedure, the virtual switch controls when the apply processes and their related jobs are ended and restarted (Related jobs are either the multithreaded database apply job (when configured) or access path maintenance (when enabled by policy and allowed by configuration).) You can start or stop other replication processes as needed on the production node.
Virtual switching uses target journal inspection processes to identify objects that have changed on the test node. Starting and stopping MIMIX also starts and stops target journal inspection. Target journal inspection processes need to remain active during virtual switch testing to avoid having a period of time where changes made on the test node are not detected. Data groups affected by the virtual switch can be started and stopped.
Virtual switch status is triggered by specific step programs in the virtual switch procedure and is reflected in the replication status of the application group and its resource groups and data groups. At the application group level, replication status will indicate statuses of virtual switch starting, virtual switch testing in progress, and virtual switch recovery in progress when no other conditions that require attention exist. While a virtual switch status is displayed, other less severe conditions can exist and you will need to drill into more detailed interfaces to see those problems. When using Assure UI portal, an important difference in the overall status column in the Application Groups portlet is that any virtual switch status takes precedence over all other possible statuses so that you can easily identify that a virtual switch is in progress. Detailed columns in that portlet will reflect any other issues that need your attention.
A failed or canceled virtual switch procedure can only be acknowledged if the failure or cancel request occurred before step MXSETVRTS. If the procedure fails or is canceled after the virtual switch starting status is displayed for replication, you will not be able to start, stop, or switch the application group until after the virtual switch procedure is resumed and runs to completion.
In interfaces for starting a virtual switch, the default value for the test node is the node identified as the first backup in the current node roles of the recovery domain. If the application group has three or more nodes in its recovery domain, you can select to use a different backup node. Replicate nodes cannot be selected as the test node.
In cascading environment, do not use the middle node of the cascade as the test node for a virtual switch.
Audits are not allowed to run during a virtual switch.
Target journal inspection will not issue notifications while virtual switch testing is in progress.
If you subscribe to MIMIX events through the Assure UI portal, be aware that some events will not be triggered during a virtual switch or when certain virtual switch states occur. There are unique events that are triggered by virtual switch status changes.For more information, see the list of MIMIX subscription events available in online help.