Performance testing tutorial – Advanced.
Jmeter Performance test scenario – Creating a Jmeter Performance test scenario
Thread Group: Thread group is the number of virtual users performing the transaction on application under test.
How to add Thread group?
- Right click on the TestPlan
- Select Add>Thread Group from the popup.
Performance testing best practices and activities.
- As first step, Performance Engineer needs to learn the requirements, performance expectations & system architecture.
Note: To learn the performance requirements & expectations, SME (Subject matter expert ) or production user can help. This information will help to define performance goals and SLAs
- Then develop best matching performance test scenarios, approach, define goals (SLAs) & sign off test plan.
Note 1: To identify the right performance test scenarios, SME can give some feedback on about the current performance issues in production / live application that he/she is facing.
Note 2: You can refer another blog of mine on different types of performance scenarios – click link load testing vs performance testing vs stress testing
- Now Performance Engineer can developer V– USER scripts and execute the scripts in Controller to generate performance analysis results
Note 1: Apart from delivering performance analysis results, Performance Engineer can coordinate with other team to monitor real time performance on Database server, Application server and also make sure load balancing is working fine while stress testing.
Some of the e.g. when performance testing is needed
- for Database upgrade,
- Application upgrade,
- Or if user complains that application performance is bad – like page response is not meeting his/her expectations.
1. Performance testing – The main purpose of Performance testing is to arrive the benchmark & baseline of the system behavior (e.g response time) under normal/moderate load. This test will help performance engineer to analyse how does the application behave under normal conditions.
2. Load testing – here load will be constantly and steadily increased over a given period of time. i.e the constant load applied for a constant period of time. For example a load of 100 users are applied for 15 min.
3. Stress test – it is meant to push the system beyond the normal load limits or extreme conditions. This test is to make sure the servers don’t crash or what point it crashes. Stress test will verify the breaking point of the system or how much load, system can withstand.
For example a load of 100 users are applied for 15 min and then 150 users load is applied for next 15 min. then 200 users load is applied for next 15 min. Verify at what point system stopped to respond.
4. Endurance testing – for easy understanding, I would say it is a “Load testing on extended period” to make sure web servers are performing longer period of time. For example a load of 100 users are applied for 48 hours
Note: Once the performance testing is completed successfully, tester can assume the performance requirements are met and he can start load testing. After that Stress testing if required.
Load testing tools in the market are
1) HPE LoadRunner is from Hewlett Packard Enterprise.
2) JMeter – Apache JMeter. It is a Java based performance testing tool works on web communication layer (HTTP)
Summary about Jmeter and performance testing
How to create Jmeter scripts?
Recording Jmeter scripts using Jmeter’s HTTP(S) Test Script recorder. There are two ways
1) Manually adding scripts ( Test plan >> Tread Group >> Add>> Sampler >> HTTP Request )
2) Recording using Jmeter’s HTTP(S) Test Script recorder
How to record Jmeter scripts using HTTP(S) Test Script recorder
Prior doing that we need to configure browser and learn about the Proxy setup
Jmeter to Application server communication need to be routed through a browser Proxy for this recording purpose. For that you may need to select the Browser (here we can use Firefox browser”
1) Navigate to Firefox >> Tools >> options >> Advanced >> Network >> Settings
2) Select Manual Proxy configuration >> HTTP Proxy = localhost & Port = 8080
How to run Jmeter in Jenkins
Jenkins setup and performance plugin setup:
1) copy performance plugin into Jenkin’s home directry before you start the test.
2) to do that – you can download the performance plugin “performance.hpi” from the site https://wiki.jenkins-ci.org/display/JENKINS/Performance+Plugin
3) how to find the home directry of jenkins – to do that you can navigate to jenkins home page (localhost:8080) and then
4) click on “Manage Jenkins” and then click on “Configure system” the you will notice “Home directy” path in there.
Configure Jmeter and create a basic script in Jmeter:
5) download Jmeter Binary zip file from Jmeter site http://jmeter.apache.org/download_jmeter.cgi
6) unzip it and go to the “bin” directory of the Jmeter folder, wherre you will notice “user.properties” file.
7) open the file in notepad or notepad++ or any other text editor
8) add a line “jmeter.save.saveservice.output_format=xml” to that. Save and close the “user.properties” file
9) create a Jmeter script and make sure you can run it in non GUI mode i.e either in DOS mode or using shell scripts in MAC (note down the commands used in respective OS and it could be used in Jenkins job setup)
Note: for windows command line run, you can use the following commands
C:\[JMETER BIN directory]\jmeter.bat -Jjmeter.save.saveservice.output_format=xml -n -t C:\[JMETER script “.jmx” file] -l Jmeter_test_results.jtl
setup Jenkin’s job to run Jmeter scripts
9) go to Jenkins home page (localhost:8080) and then click on “New Item” on “Freestyle project”
10) in the “New item” page, give some name in the item name field and go to below to section “Build” and add build step
11) add build step can be either a) “execute windows batch command” for windows or b) “Execute shell” for MAC
12) provide your command in there and save.
13) Click on “Build now” which would call the Jmeter (jmx) file and run the program
14) after the run, you can verify the Jenkins console to review the results.
Rendezvous point is Synchronization or Wait point. Rendezvous points makes Vusers to wait during test execution and multiple Vusers will simultaneously perform a task.
Rendezvous is mainly used to emulate heavy load on a particular scenario/ transaction to test the application behavior.
Latency issues generally occurs as a side effects of patch updates in windows. It may occur as audio/video streaming performance issues.
Latency issues can be tested using LatencyMon or Thesycon
Note : Thesycon is more effective on lower versions on windows like 7 or lower.
Web server serves a web content (HTML and static content) over the HTTP protocol.
Application server is a container on which one can build and expose business logic and processes to client applications through various protocols like HTTP. Some cases it will have internal web server. Application server is heavy in terms of resource usage.
Database server refers to the back-end system of a database application using client/server architecture. The database server performs tasks such as data analysis, storage, data manipulation, archiving, and other non-user specific tasks
A Web server can process HTTP requests and respond with HTML pages, provided the requests are handled by server side programs
It uses HTTP protocol to transmit the request/responses. Here Web server handled both request/response and business logic. This is more rigid design.
A web transaction using web server will be like this as below,
Browser (click operation to display some results on a grid> ==> Request goes to web server (Apache HTTP server) ==> Database