Performing General Data Optimization


You want to streamline the data as much as possible before you run it through SQL processing in a SAS Data Integration Studio job.


You can minimize the input and output overhead for the data. You can also pre-sort the data. Perform the following tasks:


Minimize Input/Output (I/O) Processing

To help minimize I/O and improve performance, you can drop unneeded columns, minimize column widths (especially from Database Management System [DBMS] tables that have wide columns), and delay the inflation of column widths until the end of your SAS Data Integration Studio flow. (Column width inflation becomes an issue when you combine multiple columns into a single column to use a key value).

Pre-Sort Data

Pre-sorting can be the most effective means to improve overall join performance. A table that participates in multiple joins on the same join key usually benefits from pre-sorting. For example, if the ACCOUNT table participates in four joins on ACCOUNT_ID, then pre-sorting the ACCOUNT table on ACCOUNT_ID helps optimize three joins. However, the overhead that is associated with sorting can degrade performance. You can sometimes achieve better performance when you subset by using the list of columns in the SELECT statement and the conditions set in the WHERE clause.
Note: Integrity constraints are automatically generated when the query target to the SQL transformation is a physical table. You can control the generation of these constraints by using a Table Loader transformation between the SQL Join transformation and its physical table.