|
|
Once you plan your load test, the next step is to perform the load testing process. Load testing is the process of identifying performance bottlenecks in your web application under normal and peak loads. This will help you tune the resources (both hardware and software) of your web application and optimize the user experience for maximum performance.
Following are the steps involved in the QEngine Load
Testing process:
Step 1: Record Transaction
Record the user scenarios or transactions of your web
application. Here, the user scenarios are the ones identified in your
planning phase in Step1.
Step 2: Parameterize Test Scripts
Once you have your basic recorded load test script, you have to parameterize the load test script to vary the input to the server. This will help you to emulate real-world testing and avoid errors arising out of duplicate values.
Identify the list of values to be parameterized. Replace the recorded values with parameters to pass a different set of data to the server for each virtual user in the load test. For example, user name and password values can be parameterized using the Dataset option in a login script to pass different user name and password for each user. The values for the parameter can be fetched from:
An external datasource (CSV or database)
A Cookie
From response text obtained for the previous request.
A hidden field
From previous URL
By executing a Javascript
Using a constant value.
From Random Data
Using Variable Store
Step 3: Associate percentage of Load
The next step is to associate the load percentage for
the recorded transactions. Depending on the number of transactions recorded
for the load test, you need to associate the appropriate load percentage.
The sum of load percentage of all the transactions in the load test should
be equal to 100%
Step 4: Simulate Load
In this step, you examine the web application's behavior under simulated load conditions. This will help you to identify whether your application is trending towards or away from its defined performance objectives. Once you identify the load type to be simulated based on the instructions given in Step 4 in 'Planning Your Load Test', select the appropriate load type and configure the workload details. For example, let us consider the following load for a sample online bookstore application:
Load Type - Normal
Number of Users - 100 users
Number of Times - 2
With this configuration, QEngine will execute the transactions
present in the load test 2 times, simulating 100 users. The test will
be completed when all the 100 users complete the transaction 2 times.
Step 5: Execute Load Test Cases
The next step is executing the load test. During load
test execution, QEngine automatically simulates the test runs for various
load types emulating real-time user roles and access patterns. It
displays runtime graphs during execution and detailed reports and graphs
after test completion.
Step 6: Analyze Test Results
The final step is to analyze the test results which
is both the most important and the most difficult part of performance
testing process. To make this process easier, the reports and graphs page
that gets displayed after load test execution are organized in a clear
and easy to understand format. This will help you to just click the various
links in the left pane and identify the performance bottlenecks. These
reports can also be viewed from the 'Reports' tab of the Web Performance
UI.
When you run the load test for the load created in Step
3, the results will be as follows:
Avg. Hits/sec or load generated against server: 20 hits per second.
Avg. Page Download Time: 4,371 milliseconds.
Avg. Throughput or Data Transfer Rate: 83,552 bytes per second.
Avg. Response Time or How Fast your Server Responds: 4,334 milliseconds response time.
Avg. Error Percentage: 25% for 120 active users when the elapsed time is at 30 seconds.
Server Monitoring Results:
% CPU Usage: 40 percent
Memory Usage: 60 percent
|
|