11 Preprocessing a Database Workload

After a workload is captured and setup of the test system is complete, the captured data must be preprocessed. Preprocessing a captured workload creates all necessary metadata for replaying the workload. This must be done once for every captured workload before they can be replayed. After the captured workload is preprocessed, it can be replayed repeatedly on a replay system.

To preprocess a captured workload, you will first need to move all captured data files from the directory where they are stored on the capture system to a directory on the instance where the preprocessing will be performed. Preprocessing is resource intensive and should be performed on a system that is:

  • Separate from the production system

  • Running the same version of Oracle Database as the replay system

For Oracle Real Application Clusters (Oracle RAC), select one database instance of the replay system for the preprocessing. This instance must have access to the captured data files that require preprocessing, which can be stored on a local or shared file system. If the capture directory path on the capture system resolves to separate physical directories in each instance, you will need to merge them into a single capture directory where the preprocessing will be performed. All directories must have the same directory tree and all files contained in each of these directories must be moved into a directory that has the same relative path to the capture directory.

Typically, you will preprocess the captured workload on the replay system. If you plan to preprocess the captured workload on a system that is separate from the replay system, you will also need to move all preprocessed data files from the directory where they are stored on the preprocessing system to a directory on the replay system after preprocessing is complete.

This chapter contains the following sections:

Tip:

Before you can preprocess a captured workload, you must first capture the workload on the production system, as described in Capturing a Database Workload .