In the following recipe, we will use FastLoad to load data from a flat file to Teradata. The following are some known issues which degrade the performance of the load utility:
- Unclean data decreases the efficiency of the apply phase
- Unclean data, like duplicate unique indexes, duplicate rows, and constraint violations put loads on error tables
- Uneven distribution of rows
The loading of data is done in staging tables when following the traditional ETL process. We can club FastLoad with SQLs to increase the performance of ETL, as shown in the following figure:
Fastload can be invoked on a channel attached system to load data via a disk or using a flat file on a network attached system. You can load data from:
- Special input module (INMOD) routines
- Any other device providing properly formatted source data
