I want to make some simple load tests for a web application - i want to do some benchmark-tests.
**how to do this: **I remember one of the tools I have seen in a stage and I liked a lot was Selenium, the famous tool of the past. On the downside of Selenium - it is intended for developing use cases for testing user interface of web application. My needs are much much simpler, simulate concurrent users. i have done some searches: I have seen Apache Benchmark ‘ab’ but one thing I would like to do is simulate a scenario where we begin with 5 users …
I have heard about JMeter ( http://jmeter.apache.org/ ). A fairly nice tool - but not as simple as ab from Apache or httperf the famos tool, but i guess that JMeter
it isn’t a too complicated tool. JMeter can be used for almost every server benchmart tests. In the JMeter docs we can see a simple tutorial for doing web load tests. ( cf http://jmeter.apache.org/usermanual/build-web-test-plan.html ).
Another tool is curl-loader which supports customed number of users per-second and test different URLs.
A third tool is named siege, which is quite similar to ab(apache benchmark)
**conclusio: **are there some other tools - eg some that are more simple than the above mentioned!?
Be aware that simple tools may provide simple results valid <only> for certain measurements (like general hardware capacity).
The earliest tools simply recursively queried a page (eg GET) and tabulated how many you could hit within a certain amount of time.
As hardware became more powerful, solutions deployed multiple client machines to apply a heavier load.
But, then people noticed that this kind of testing results only in artificial and unrealistic values (In a real world, would everyone hammer your home page and click on nothing else?). Any click/action on a page would generate a different load, and if is the action’s first time then the results aren’t in the webserver cache. To build the content, there might be calls to a database, retrieving varying amounts of data perhaps even related to the remote client.
So, “better” website benchmarking is done nowadays by constructing scripts that emulate real User activity, and may even be derived from actual web server logs.
That said, nowadays there is a big “back to the future” where a large number of websites today are not complex and are static or Single Page Applications which easily fit within memory caches and don’t require a database backend.
So, YMMV.
Select and choose your benchmarking tools to return results valid for what you are doing.
