Update a data mart

You can refresh the data in a data mart. You can update the data mart definition when you change the warehouse schema.

If you plan to use partial data refresh for data marts, including trickle feed, the value of the MAX_PDQPRIORITY configuration parameter must be set to 50 or greater during the initial data load and partial data refresh.

You can use different methods to update data marts depending on your needs. For best results, reload the entire data mart on a regular schedule, for example, once a week.

Table 1. Methods for refreshing data marts
Task Method
Continuously refresh the data in the data mart Start trickle feed to refresh the data in the data mart as data is inserted in the database. You specify how frequently the data mart is refreshed.

run the ifx_setupTrickleFeed() function.

Important: You must periodically disable trickle feed and reload the data mart. Trickle feed stops running after 32000 refreshes.
Refresh the data in the data mart when you choose Refresh the partitions in the data mart. Dimension and fact tables that have stale data are updated.

run the ifx_refreshMart() function.

Refresh a specific partition or table when you choose Drop the specified partition or table and then reload the data.

Run the ifx_dropPartMart() function and then the ifx_loadPartMart() function.

Reload all the data in the data mart

Reload the data in your data marts once a week to ensure that they are accurate.

Disable the data mart and then load the data. Any attached or detached fragments for dimension or fact tables are added or dropped. Query acceleration is temporarily disabled.
Update the data mart definition by re-creating the data mart Drop the data mart and then re-create the data mart with the updated data mart definition. Query acceleration is temporarily disabled.
Update the data mart definition by replacing the data mart Create a new data mart with an updated data mart definition, load the data, and then drop the original data mart. Query acceleration continues without interruption, however, you need enough main memory and disk space for two data marts.
Important: There is a 32 KB limit on the number of refreshes from Informix® to Informix Warehouse Accelerator before a reload is required. If you exceed the limit, an alert_type ERROR with alert_color YELLOW is added to the sysadmin:ph_alert table indicating the limit is reached. Here are some examples of trickle feed frequency and the number of days before reload is required:
Frequency Number of days before reload
5 minutes 113 days
15 minutes 341 days
60 minutes 1365 days