Hi all,
I am a new JMeter user and would appreciate some advice on how to go
about achieving my slightly unusual goals.
I want to drive the configuration of JMeter dynamically in
conjunction with an evolutionary algorithm which attempts to find the
most stressful configuration of JMeter (within certain constraints of
course) for my server.
My code will consist of the evolutionary algorithm which will decide
upon and tune the JMeter configuration, including:
- Number of Thread groups
- Looping and User characteristics of those groups
- Arrival times through addition and manipulation of timers
- Deciding which resources to visit
- etc. - pretty much most of the components JMeter exposes will be
configurable by my evolutionary algorithm.
This evolutionary algorithm needs feedback on how it is doing
however, which JMeter provides for in the form of listeners.
I am seeking advice on the implementation details of the above, for
example, I could write the evolutionary algorithm part outside of
JMeter entirely - just have it generate XML descriptions of a
configuration, and read the output from the log for feedback. This
will certainly provide the loop of EA_Generates_Configuration-
>JMeter_Input->JMeter_Runs->JMeter_Output->Feedback_Tunes_EA...
However, I am wondering whether the JMeter plugin architecture might
be better/faster/more flexible for what I am trying to do. Or perhaps
there is another approach (hybrid?) which you might suggest.
Thanks for your time and advice, it's much appreciated and I'm sure
will help keep me from flopping around too much at the beginning of
this (non-commercial academic research) project.
- Peter
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
