History recovery

History recovery enables you to recover archive data for time periods when the interface was not running or otherwise unable to collect data. During history recovery, the interface updates target PI points with values from the source archive, according to the settings you configure.

The interface performs history recovery after the following events:

• During interface startup (before starting real-time data collection)

• After restoring a lost Data Archive connection

• During queue overflow error recovery

• Whenever a new target point is added to the interface

• Interface failover

  When failover is triggered and the backup interface takes over as primary, the starting point for history recovery is set on a tag-by-tag basis. If the current value for a tag is more recent than the starting point of the recovery period, history recovery begins at the tag’s current value.

You can specify the recovery period, time increments within the total recovery period, a pause time between increments, and the maximum number of archive events that are returned per data request. These parameters enable you to manage the performance impact on the source and target PI Data Archives by throttling the data collection rate.

Time-range history recovery is configured in the interface startup configuration file, which you edit using PI ICU. When time-range recovery is enabled, the interface starts up, recovers archive data for the specified start and end time then exits. Specify the recovery start and end times in the time zone where the interface runs, even if the source Data Archive is in a different time zone than the machine where the interface runs. Ensure that the target archive has enough space for the data to be recovered.

The starting point for history recovery is determined on a point-by-point basis. If the current value for a point is more recent than the starting point of the recovery period, history recovery begins at the point’s current value, to prevent data overlap, streamline data retrieval, and prevent out-of-order data. For points that collect future data and newly-created points with no archive data, the interface performs history recovery for the total recovery period.

History recovery is performed independently for each scan class. Source data is written to the target archive in one of three modes: append, replace or no replace. This setting is configured on a point-by-point basis. The interface logs the progress of recovery.

By default, during history recovery, the interface also collects real-time exception data every five seconds. To configure this time interval using PI ICU, go to the Optional tab and select the check box Set time interval between clearing exception queue during history recovery. As it nears the end of its point list, more points are in the update list and the time to collect exception updates increases. This extra overhead can significantly increase the time required to complete history recovery.

By default, history recovery bypasses compression testing on the target server when its version is Data Archive 3.3 or later. Target Data Archives that are pre-3.3 always apply the compression-test to incoming data.