Input Recovery: RDBMS values to PI Data Archive

To copy bulk data from RDBMS tables to PI, you can run the interface in the recovery mode specifying the recovery start time. To ensure that system resources (CPU and memory) are not taxed by queries that retrieve potentially huge amounts of historical data, it is recommended that you specify the Input Recovery Step to let the interface read data in chunks rather than return back one (huge) result set. The following example shows the principle:

SELECT timestamp, value, 0
FROM table1
WHERE timestamp > ? AND timestamp <= ?
ORDER BY timestamp ASC; P1=TS P2=TE

If the Recovery End Time is left empty, the interface processes the period since the Recovery Start Time till current time and then continues in normal data collection. If the Recovery End Time is populated, the interface processes the interval (for all input points) end then exits.

Note: During recovery, provided the Recovery Step is specified, the TS and TE placeholders are dynamically calculated. If the interface continues running after it finished the recovery, the TE placeholder is populated by current time.

The next figure shows how the recovery proceeds applying the recovery step.