International Journal of Trend in Scientific Research and Development (IJTSRD)

Volume 4 Issue 2, February 2020 Available Online: www.ijtsrd.com e-ISSN: 2456 – 6470

Load Runner: Methodology to Performance Testing

Ashish Kumar
Performance Engineer, Tata Consultancy Services Ltd, Mumbai, Maharashtra, India
Email: ashish.ksinha@yahoo.co.in / ashish.k40@tcs.com

ABSTRACT How to cite this paper: Ashish Kumar

The paper covers the most important part of testing which is essential for "Load Runner: Methodology to
testing the performance of application before going ‘Live’. In my research and Performance
analysis with Microfocus owned tool ‘Load Runner’, we will discuss about it Testing" Published
deals with web based application and approach to calculate transaction per in International
hour (TPH) for test execution. As tool supports many protocols based on the Journal of Trend in
nature of application. Performance testing is used to analyze the real time Scientific Research
response time for business transaction. Application be constant with and Development
increasing load or with simultaneous users should not affect the performance (ijtsrd), ISSN: 2456- IJTSRD30155
of the application, is our main motto as a Performance Tester/Engineer. 6470, Volume-4 |
Issue-2, February 2020, pp.878-880, URL:
KEYWORDS: Performance Testing, Load Runner, Performance Center, Virtual www.ijtsrd.com/papers/ijtsrd30155.pdf
Generator, Transaction, Load Runner Analysis
Copyright © 2019 by author(s) and
International Journal of Trend in Scientific
Research and Development Journal. This
is an Open Access
article distributed
under the terms of
the Creative Commons Attribution
License (CC BY 4.0)
(http://creativecommons.org/licenses/by
/4.0)
I. INRODUCTION
Application Performance Testing is all about the Performance testing come in light when some unexpected
performance and quality of the application. The performance cause start arising even after Manual or functional testing.
testing will insure the behavior of application under their Many companies loses its credibility as the unexpected
expected workload. Performance testing is a type of Non- behavior of application come in picture in production
functional testing, which simplifies that these testing has no environment when sudden increase in real time user, end
relation with the functionality of the application. The user, back office or admin hits the application. To overcome
performance testing is done for a web application for its this big challenge, Performance testing play a vital role. It is a
speed, scalability and stability (S^3).The goal of performance tool based testing where the tool records the events of
testing is not to find bug and fix them; it is moreover for request and re-play it with simulation of different users, just
elimination of performance bottlenecks. say real time virtual user.

II. TYPES OF SOFTWARE TESTING

Fig1: Types of Software Testing Fig2: Types of Performance Testing

 Load Testing – In load, testing application has been

checked under various load to test any bottleneck of the
application.

@ IJTSRD | Unique Paper ID – IJTSRD30155 | Volume – 4 | Issue – 2 | January-February 2020 Page 878
 International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
 Stress Testing – In stress testing, web application is In addition, to reduce the high count of engineers to perform
checked under extreme workload to see the breaking manual testing, Load testing gives the entire solution for
point of an application. performing this type of testing.
 Scalability Testing – It is used to check the load with
increasing number of user of an application, and it is
helpful in increasing the capacity of software system.
 Volume Testing – In volume testing, large number of
data is stored in database and overall web application
behavior is observed. It checks the correctness of an
application under varying database volume.
 Spike Testing – When there is sudden large spikes in
the load generated by users, then spike testing is done.

III. Software Testing Tool – Load Runner

Load Runner is software-testing tool owned by Micro Focus.
Load Runner is a best tool as per market survey, which
Fig4: How Load Runner works
simulates user activity by generating messages between
application components rather than simulating interactions IV. Automated Performance Testing
with the user interface such as keypresses or mouse With the direct approach to perform the test using Load
movements. Runner, you can optimize resources, best suggestion for
hardware configuration, and the best network requirement
HP acquired Load Runner as part of its acquisition of
based on application architecture, and debugging the
Mercury Interactive in November 2006. In Sept 2016,
transaction response time based on agreed service level
Hewlett Packard Enterprise announced it is selling its
agreement (SLA’s).
software business, including Mercury products, to Micro
Focus. As of 01-Sept-2017, the acquisition was complete.  Planning & Identification of Key business Area in
Performance Testing
Components of Load Runner
A user requirement is translated to performance objective
 Virtual Generator
for observing realistic behavior of application. In planning
 Analysis
and drafting the performance test plan, we need to
 Controller
understand the key business area and their impact on
 Load Agents
complete business flow of particular LOB based application.
 Agents Process
Test plan need to complete with the current and projected
 Recording
user specified objectives, which comprises the current
statistics of production. For designing the plan, we must
understand the architecture of application and we should
have strong knowledge to compare the production details
for apple-to-apple comparison.

 Defining the Data

You need to define what data is required onwards to
perform the test. The data can be created dynamically, or
making a clone of master user, or writing few interesting
creation query in Database. By using realistic data/user, you
can create accurate load test.
Fig3: Components of Load Runner
 Defining Testing Approach
When you start the recording in Vugen for a new application. You should have depth knowledge in creating the strategy
Record the same business flow twice that will help in for testing applications. You can choose any type of
comparing both vugen scripts and an easy way to find out performance testing, say as Load, Stress, Capacity, Volume,
the dynamic values. Once we a list of dynamic values, we can Spike, etc based on the user requirement. This type of testing
easily start correlation. Correlation in Load Runner is can be performed with Load Runner.
defined in two ways as automatic and manual correlation.
 Making of Performance Scripts
Suppose we had a dynamic value ‘ABC001”, when you are When you are good with the above steps, we need to record,
validating the same in code generation, you will have Left enhance the scripts for completed flow. You need to write
and Right boundary e.g. 78_LR_ABC001-XX. In this example, lr_start_transaction (); before the particular web request and
70_LR_ will be left boundary and _XX will be right boundary. lr_end_transaction (); after completion of that request. This
Any dynamic value in this boundary will be captured function will count the response time of web request for
through tool. transaction, which will be helpful in comparing the same
with agreed SLA’s. For enhancing the script, any value you
 Why Load Runner? enter via keyboard at the time of recording needs to be
Traditional way of testing approach is known as Manual parametrize and any dynamic value you observed in code is
testing which offers partial solution to load testing. When to be correlated.
Client/Product owner wants to ensure that what load their
application can sustain, load based testing came into picture.  Test Scenario and Execution

@ IJTSRD | Unique Paper ID – IJTSRD30155 | Volume – 4 | Issue – 2 | January-February 2020 Page 879
 International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
Once the test script is ready and validated with few
iterations (Say 5) and different user, script should be On dividing the result with 3600, we will get the response in
uploaded in Performance center / controller. seconds (TPS). Remember TPH means transaction per hour.
We need the make the scenario for our test executions. The numerical explanation for little law is explained as
 Dedicated Load generator is to mapped according to the below in spreadsheet formula
number of Vusers which you are going to test.
 While making the scenario, ramp up and ramp down
should be given and a steady state of run must be
mention as per requirement.
 Monitor varieties of graph while running the execution

We need to enable run time settings with our requirement.

1. TPH (Transaction Per Hour)
Little law states the formula as below Many performance engineer/tester believes whatever the
Number of Vusers = TPH * (Pacing + Think time) / 36000 result they get is accurate, but it does not be!
2. Logs
For example, if you run a Load Runner test of 1000
When we run the test in performance center, we need to
concurrent users, Load Runner will give some results. Never
disable the full logs. Why? Because it will write the logs
assume the given result is 100% accurate. Always it is
which will require memory. In addition, if your test is of
recommended to cross verify the results using the formula
higher virtual users, it will write full logs. So recommended
stated above.
way is to disable the full logs and write the logs only if error
occurs.
Let us say, TPH is 50 and Avg. response time including think
3. Pacing/Think Time time is 15 seconds.
With the help of little law, calculate the pacing/ think time
desired for your execution Number of users = 50*15 = 750
Rest other options will be treated as per the project
However, our expected result was 1000, something is wrong
requirement.
here!  So we need to place the count with the formula stated
above (refer spreadsheet picture above) to calculate the
 Report And Analysis
accurate measurement.
Once the execution is completed, download Load Runner
analysis (.lra) file from PC. Open LRA file in Analysis
Reference:
component of LR. There you can see some pre-defined
[1] https://en.wikipedia.org/wiki/Load Runner
graphs, and you can select more graph based on your
client/project requirement. It is recommended that select [2] https://www.softwaretestinghelp.com/Load
some customized report template, and add flavors based on Runner/Load Runner-scripting-challenges/
your requirement and generate the report.
[3] https://www.softwaretestinghelp.com/Load Runner/hp-
Load Runner-tutorial-8/
V. TRADITIONAL APPROACH TO CALCULATE
FORMULA [4] https://www.perftesting.co.uk/performance-center-
Little Law: architecture-overview/2011/08/24/
Little’s law named after “John Dutton Conant Little” who was
Renault Professor at the MIT. It states the number of request [5] https://perfmatrix.blogspot.com/p/hp-Load
Runner.html
in system (closed) is equal to the product of average number
of Requests serviced per unit time and the average time each [6] https://www.guru99.com/understanding-vugen-in-Load
Request stays in the system. Runner.html

Numerically, [7] https://www.softwarehour.com/Load

N = (R + ThinkTime) * X Runner/parameterization-Load Runner-step-step/
[8] https://www.tutorialspoint.com/Load
Where,
Runner_online_training/Load
N = average number of users in a system
Runner_parameterization.asp
X = average throughput or departure rate of users
R = average time spent in the system or response time [9] http://www.geekinterview.com/question_details/15651

VI. EXPERIMENTAL APPROACH [10] http://nonfunctionaltestingtools

Based on the little law approach, we had .blogspot.com/2015/01/verifica
Average Number of users = (Average Time + ThinkTime) * tions-and-error-handling-
Average Throughput in.html
Additionally, we can talk more about simplifying this Mr. Ashish Kumar, an engineering
formula with all other timing required to manipulate the graduate from VelTech University,
formula is think time, pacing, average response time with 1 Chennai in the field of Electronics &
iteration, etc Communication Engineering. After
completing his degree, he worked in
No. of users = {(Average Response Time in 1 iteration +
IT companies like HCL, Accenture and currently working
Pacing Time + Think Time)* TPH}

@ IJTSRD | Unique Paper ID – IJTSRD30155 | Volume – 4 | Issue – 2 | January-February 2020 Page 880
 International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
with TCS on the role of Performance Engineer. He has expert
knowledge in performance testing tool Load Runner with
multiple hands on experience in monitoring tools.

@ IJTSRD | Unique Paper ID – IJTSRD30155 | Volume – 4 | Issue – 2 | January-February 2020 Page 881