In very simple terms (and I do mean very simple) if you have a graph showing
response times over time then 'good' means level, steady response times that
do not change; 'bad' is when they go up. This is pretty simple, right? The
problem a lot of people new to load testing have is that they do not set
proper objectives so they are unable to decide if they should stamp things
with a PASS or a FAIL. (See rule numero uno in my last post.)

Think about it. If you run a load test with 1 user making 1 request every
minute then the response times will most likely always be the same. If you
run a test with 2 million users making 500,000 requests per second then,
probably, you will see response times rather high, and a fair few errors to
boot.

It's not that hard really, just decide on the point inbetween those two load
levels, and then *stick to it*. Use a Constant Throughput Timer to do this.
Seriously, I really mean it, use this control.

How you decide on the level that is right for your project requires talking
to people. Tip: The trick is to make an estimate of the peak volume of
traffic this site will actually get after it goes live. Don't underestimate
this step; most sites either fail post launch or are needlessly delayed
because someone got this wrong.

There you go, performance testing in four paragraphs.

-----
http://www.http503.com/
--
View this message in context: 
http://jmeter.512774.n5.nabble.com/Timer-how-to-delay-the-HTTP-request-in-the-thread-tp5488599p5500374.html
Sent from the JMeter - User mailing list archive at Nabble.com.

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to