In my organisation , application being migrated to on-premise site to Azure (SaaS, PaSS), and the current on-premise JMeter setup does needs replacement as well.
So Azure load test looks suitable choice , but doing the setup the internal / private end point looks complicated (Vnet injection method). So I am wondering if someone can share their experience of setting up any alternate tools in Azure to conduct testing of internal applications.
Running the test from within Azure does sound like the most logical solution. A basic option is getting set up with a Linux instance with a decent amount of CPU and RAM to match the test, then manually SSH to the instance, install JMeter and continue from there.
However, my preference would be to set up some pipeline or process to automatically provision an instance, install JMeter and the test, run the test, publish the results and drop the resources. A bit more to set up, but a lot easier to work with and much cheaper hosting costs.
You could also use Azure container apps, again provision a JMeter container and run the test from there.
In terms of tools though, I would stay with JMeter. It works well in Azure, you already have the scripts/data/etc, and moving to other tools (for example K6) may not give you any additional benefits.
Thanks Simon. I was thinking in the similar line however i was not sure if Microsoft would not give all necessary bandwidth between my Linux / Container apps Jmeter and application under test compared to what they offer with Azure load test. However if this not true and someone here have experienced doing this recently then i am more confident to try this out. Also, if anyone have done then, please share any reference / guide available to try this out in Azure. Much appreciated the help. Thanks
The bandwidth between apps inside Azure is basically unlimited, the only limit is on the VM config itself. You might need to move up a VM category if you find you’ve not got enough bandwidth on the Jmeter instance. Things can get a bit messy when either JMeter or the SUT is outside of Azure, and it gets even messier if you’ve got one on Azure and one on AWS (for example). Fortunately you’ve got the easiest set up.
We started with the most basic VM, something like 2vcpu and 16Gb of memory, and found it covered everything we could possibly throw at it. Provision a small VM, benchmark it based on your tests, then upgrade if necessary. You’re in the cloud, take advantage of the flexibility.