After the data has been loaded and the database optimized, the typical first step in running an evaluation is to run some queries in a single-user mode of operation.

This can be done either by using a BI tool (typically from a remote machine) or by using scripts (typically run locally on the VectorH master node). Both have advantages and disadvantages. Ultimately, if a BI tool is what the eventual end-users will be using, then some testing needs to be undertaken with the target tool. However, running some queries directly against the database without using a BI tool can help to eliminate network effects from the performance of a query, and so provide a clearer view of where the overall execution time is being spent.

In addition, it is easier to set up concurrent user testing later if you have the target queries already scripted.

If you want to test using scripted queries, a tool called “runall” is available for download on the Actian Github site (see Reference 7 (References and Further Information)) that will automatically run a series of SQL statements in script files and provide timing information. This tool can also be used to execute concurrent user testing.

After running these initial query tests and evaluating the performance results, look to optimize the schema and queries (if needed) to get the best possible results. The next section describes the main areas to consider in optimizing performance.