0


Perhaps its a simple one but I can't find any leads.
We develop technologies for the server itself (not the app-level of the server but technologies for the HW levels),
For our testing, we need to simulate a stress test of a real-usage scenario of a datacenter / Internet farm server (the test server runs SQL , apache/asp, ftp)

All solutions I found on the net are aimed at cases in which you have your own end-user server application (on top of SQL / WEB), so they just flood it with requests.

We need to diagnose the server at real life stress conditions... writing an app that will overload cpu / disk won't do. neither just a SQL loop
Not to mention that even if we will wrote a full blown app just for the testing, it's still one app, while a server in a farm / datacenter will server hundreds of different apps.

Is there a free full-blown web app w/ sql overload and web that one can jmeter on top of it?
Any other Ideas?

Thanks in advance!

Jarvis80
  • 13
  • 4

1 Answers1

0

Best way to do this is to have mixed load workflows in terms of SQL requests and run them. You need to know the flow/sequence of DB queries for your application. You can get these details by enabling logging for queries.

Once you have figured the mix of queries use JDBC sampler in JMeter to load test DB. Use timers between JDBC samplers so that you are not continuously loading DB rather mimic the real-life scenario.

I hope this one helps!

rachna bafna
  • 501
  • 3
  • 7
  • Hi, Thanks, but that assumes we have a DB/App, which we don't have. for confidentiality issues I cannot elaborate in specifics, but for the discussion, lets assume we're developing a faster hard drive for servers, aside from engineering-based benchmarking, marketing needs "real-world" usage benchmarking, and we cannot simply take the drive and plug it in a data-center since its a prototype... no one will take the chance... Even if we'll develop a test app, it will never be authentic and close to real-world scenario. full-blown overloading is easy, but not necessarily reflecting – Jarvis80 Dec 18 '15 at 10:04
  • If its hard drive with faster retrieval in that case you might want to measure it in terms of disk latency, and IOs the benchmark should be along the similar lines I believe. – rachna bafna Dec 19 '15 at 22:56