Testing for Scalability - Expert Oracle RAC Performance Diagnostics and Tuning - page 106

Workshop
In this workshop we discuss the various steps involved in the burnout testing of the application. Project time lines
should not affect the scope of this testing and this phase of the testing should not be ignored. There are several real-
life failure situations that could be quoted because this phase of the test was neglected.

Step 1
Ensure the all business real life scenarios of the applications have been captured using a simulation tool such as
LoadRunner. All functional areas of the application included in this phase of the tests should be documented.
The user simulation, which includes think time and number of concurrent sessions, functional areas of the
applications,and so forth, should be similar to a production workload.
Tests should be configured to execute continuously for 40-60 hours without interruption. Any interruption would
require the tests to be started all over again and repeated for another 40-60 hours.

Step 2
All tests and stages of the tests should be recorded using a template similar to the one illustrated in Table 4-11 . How
do we record a test that is scheduled for 40-60 hours of run? Test results for long-running tests can be recorded at
various checkpoint intervals. Checkpointsshould be at a consistent time interval to measure the various performance
characteristics of the application, for example, CPU consumption of the database server and application servers,
response times for the various functional areas, network and throughput,and data transfer rates between primary and
the disaster recovery sites.Keeping the same interval to capture a checkpoint matrix of the environment will help in
easy comparisons and in final reporting.

Table 4-11. RAP Phase VIII Recorder

Checkpoint # Checkpoint

Date

Checkpoint
Time

# of Users Checkpoint/Verified
Component

# of Db
Servers

# of App
Servers

Description Status

Step 3
In our workshop, the total duration of the test is planned for 48 hours; and to have an even interval, we collect a
snapshot of the environment every 4 hours (see Table 4-12 ).

Table 4-12. RAP Phase VIII Recording Initial Test

Checkpoint # Checkpoint

Date

Checkpoint
Time

# of Users Checkpoint/

Verified
Component

# of Db
Servers

# of App
Servers

Description Status

1 02/16/2011 8:00 am 100 N/A 6 8 Start of RAP

VIII

Success

2 02/16/2011 12:00 noon 100 Entire stack 6 8 Overall

health of the
environment

Good

Note: N/A = not applicable.