Jmeter, what can not you do with it?
May 14, 2014
Jmeter, what can not you do with it?
Score 9 out of 10
Vetted Review
Verified User
Software Version
2.9
Overall Satisfaction with JMeter
Jmeter is used by our group in Intuit to do performance tests and analysis for all the different services that are part of the division. It is slowly and surely becoming a common tool for all the groups in Intuit primarily due to the ease of use and the no cost attached to it. Jmeter helps us to construct tests in minutes and the scripts can be run from any platform which has Java installed on it. Jmeter test plans (scripts) can be shared with the product development team to run on their local or dev environment to get performance insight at an early stage in the lifecycle of the product. Jmeter tests can also be run from a command line which makes it easy to make it part of Jenkins job for CI/CD framework.
- Easy to construct Webservices script - You just need to add a sampler with the endpoint, payload and header information. And boom you can run a load test on the service with varying load and capture the response time and throughput. In a commercial tool like Loadrunner or Silkperformer, this process would take at least couple of hours, while with Jmeter we can do this in 10 minutes or less.
- Modularizing the scripts - Jmeter has a very good way of organizing the scripts and the reusability of the existing script (transactions) is also as easy as copy pasting the script.
- Jmeter tests can be run from anywhere, so you can run it from the idle webserver itself if you have some firewall restrictions to test it from your local or other environment.
- Jmeter distributed tests can be spawn from the AWS and results collated to see the performance from different regions in the world.
- The most not so good thing about Jmeter comes with the distributed testing. When you setup a distributed testing say with two machines, the load being generated from each machine has to be the same. So we can not separate the test plan and say run these x scripts on machine 1 and y scripts on machine 2. Both machines will run the same load with all the scripts. There is a standard feature available in all the commercial tools to specify which scripts and how many Vusers to be run from which machine in a distributed testing.
- Reporting and graph visualization aren't that great. The option of adding external plugins for reporting does enhance it to some extent but not anywhere what we can get from Industry standard tools
- Jmeter for UI based scripting doesn't work easily as the websites offload more work into their Javascript in the UI. Jmeter does not execute the Javascript and we may not be able to measure the real user experience.
- Faster time to construct Performance Tests
- Ability to run tests from anywhere
- Include the performance tests in the CI/CD framework
I have used LoadRunner and Silkperformer, and so far Jmeter turns out be the easiest to use of all these. While each of them have their own ROI, Jmeter can be picked by anyone in hours and start testing within a day. While with other tools, we need to get license, install them (takes a while) and setup tests and firewalls, etc.