There are five stages when
performance testing HTTP applications: test creation, test editing,
workload emulation with schedules, schedule execution, and evaluation
of results.
- Test creation. Although it is possible to write a test
from scratch, you generally create HTTP performance tests by recording
representative interactions with an application. These actions are
saved and a test is generated from these recorded actions. Test recording
and related tasks are explained in Recording HTTP tests.
- Test editing. After recording a test, run it individually
and inspect the results to make sure that the HTTP tests are doing
what you expect. You might want to edit your tests. Typical changes
that you might make are:
- Workload emulation with schedules. After editing HTTP tests,
you create a schedule. You add user groups to
the schedule and add appropriate tests to each group to emulate a
task. A typical schedule contains the following:
- User groups and tests. A schedule requires at least these items
to run.
- User groups running from a remote location. This separates your
workbench activity and your load-adding activity. For more information,
see Running a user group at a remote location.
- Optional schedule items, such as loops, delays, and think time
behavior settings, to further emulate a load. For more information,
see Emulating workloads.
- Schedule execution. You typically run a schedule as explained
in Running a local schedule or test.
- Evaluation of results. Reports are displayed during the
schedule run. You can also regenerate reports after the run, customize
reports, and export reports in HTML format, so that others who do
not have Performance Tester can see them. For more information, see Evaluating results in the Eclipse viewer.