Incremental Aggregation is the process of capturing the changes in the source and calculating the aggregations in a session. This process makes the integration service to update the target incrementally and avoids the process of calculating the aggregations on the entire source. Consider the below sales table as an example and see how the incremental aggregation works.
For simplicity, I have used only the year and price columns of sales table. We need to do aggregation and find the total price in each year.
When you run the session for the first time using the incremental aggregation, then integration service process the entire source and stores the data in two file, index and data file. The integration service creates the files in the cache directory specified in the aggregator transformation properties.
After the aggregation, the target table will have the below data. Target:
YEAR PRICE ---------- 2010 600 2011 1100 2012 700
Now assume that the next day few more rows are added into the source table. Source:
Now for the second run, you have to pass only the new data changes to the incremental aggregation. So, the source will contain the last four records. The incremental aggregation uses the data stored in the cache and calculates the aggregation. Once the aggregation is done, the integration service writes the changes to the target and the cache. The target table will contains the below data. Target:
Points to remember: 1. When you use incremental aggregation, first time you have to run the session with complete source data and in the subsequent runs you have to pass only the changes in the source data. 2. Use incremental aggregation only if the target is not going to change significantly. If the incremental aggregation process changes more than half of the data in target, then the session performance many not benefit. In this case go for normal aggregation.
Note: The integration service creates a new aggregate cache when A new version of mapping is saved Configure the session to reinitialize the aggregate cache Moving or deleting the aggregate files Decreasing the number of partitions
Configuring the mapping for incremental aggregation:
Before enabling the incremental aggregation option, make sure that you capture the changes in the source data. You can use look-up transformation or stored procedure transformation to remove the data which is already processed. You can also create a trigger on the source database and can read only the source changes in the mapping.