Preparing Your Application for Tracing
To measure performance, you trace an application over a period of time or at specific points. Before OPE can do this for you, you have to take these steps:
You must specify through control data how OPE is started (start mode). Optionally, you can specify how tracing is performed (buffer size, environment, flush, event, query, process identification number, and start command), and where the collected data is to be stored (trace file).
- Specify your OPE control data as environment variables.
Tracing is started automatically when the OS_OPE_TRACE environment variable is set to auto, or manually when the OS_OPE_TRACE environment variable is set to manual, and an os_ope::start call is encountered.
This step is optional. It is necessary if you want to trace only selected parts of your application's processing or if you want to get more information than OPE provides by default.
- Modify your application so that it can call OPE-specific methods from the OPE Retriever API.
The appropriate OPE Retriever is loaded dynamically when ObjectStore encounters a valid specification of the OS_OPE_TRACE environment variable.
- Run your application.
You can run your application and, at the same time, let OPE measure its performance and create a trace file. From the trace file you create a performance database. You can view this performance database as diagrams that show you OPE's measurements in graphic and textual form.
OPE must write a complete trace file before you can create a performance database and view the first results of tracing. For automatic tracing, the application trace is created after the application finishes processing. For manual tracing, this is done after the application stops tracing by using the specified methods of the OPE Retriever API.