Guidelines for processing rows in staging tables
WebSphere Commerce Analyzer considers the following factors when determining when a row should be processed from a staging table:
- The type of query being performed
- The value of the IBMSNAP_OPERATION column.
- The value of the IBMSNAP_LOGMARKER column.
- The value of WebSphere Commerce Analyzer EXTRACTION_COUNT parameter.
- The value of WebSphere Commerce Analyzer WSA_EXTRACTION_COUNT parameter .
- The value of WebSphere Commerce Analyzer TIME_CUT_OFF parameter. TIME_CUT_OFF is set at the beginning of the replication cycle.
- The value of WebSphere Commerce Analyzer TIME_CUT_OFF_PREV parameter. TIME_CUT_OFF_PREV is the beginning of the last successful replication cycle.
- The value of WebSphere Commerce Analyzer TIME_WINDOW parameter, which is the time between the last extraction and the beginning of the last replication cycle.
- The value of WebSphere Commerce Analyzer WSA_TIME_CUT_OFF parameter.
- The value of WebSphere Commerce Analyzer WSA_TIME_CUT_OFF_PREV parameter.
These factors are taken into account by calling the following WebSphere Commerce Analyzer user-defined functions:
IWHWSA.PROCESS_ROW
(OPERATION, IBMSNAP_OPERATION, LOAD_STATUS, IBMSNAP_LOGMARKER)Where:
- OPERATION
- A
- looking for rows which have been inserted or updated during the current time window.
- D
- looking for rows within the extraction-time window which have inserted previously into the WebSphere Commerce Analyzer data mart but have recently been deleted.
- I
- looking for rows within the extraction-time window that have been inserted into the WebSphere Commerce Analyzer data mart.
- J
- looking for rows with IBMSNAP_LOGMARKER greater than the TIME_CUT_OFF_PREV that have been inserted into the WebSphere Commerce Analyzer data mart.
- L
- looking for rows within the extraction-time window regardless of whether they have been processed.
- N
- looking for rows which have not been inserted into the WebSphere Commerce Analyzer data mart regardless of whether they are within the extraction-time window.
- U
- looking for rows within the extraction-time window which have been inserted previously into the WebSphere Commerce Analyzer data mart but have since been updated.
- V
- looking for rows with IBMSNAP_LOGMARKER greater than the TIME_CUT_OFF_PREV that have been inserted previously into the WebSphere Commerce Analyzer data mart but have since been updated.
- IBMSNAP_OPERATION
- D
- Replication-specific variable indicating this row was deleted
- I
- Replication-specific variable indicating this row was inserted
- U
- Replication-specific variable indicating this row was updated
- LOAD_STATUS
- An integer that indicates the last time this row was processed by any WebSphere Commerce Analyzer ETL operation. A -1 indicates that this row either is not processed or has been processed but LOAD_STATUS is not yet updated. Whenever a row from a staging table is processed, the LOAD_STATUS column is updated with the value of the extraction count parameter in the parameter table. Extraction count is simply a counter that is incremented each time a WebSphere Commerce Analyzer extraction cycle is started.
- IBMSNAP_LOGMARKER
- The source-specific time stamp when the operation specified in IBMSNAP_OPERATION took place. The one exception to this is the first time a table is replicated; in that case, it is the source-specific time stamp during which the replication took place. This column is compared to the extraction-time window for the specific source.
(C) Copyright IBM Corporation 1996, 2006. All Rights Reserved.