Skip navigation.
Home » blogs » Alexander Podelko's blog

Open Source Load Testing Tools

Submitted by Alexander Podelko on Tue, 12/10/2004 - 14:48. OpenSTA | performance testing
I am looking at OpenSTA as an additional tool (to LoadRunner) for our company. The tool really looks quite professional and solid, although many minor problems (not to mention factual lack of analysis part) annoy and make its using questionable when you have other good choices.

I have a feeling that there is some kind of decay in the OpenSTA community (www.opensta.org). All kind of activities are down, no releases last 1.5 years... I wonder what is going on? As a tool improves more people should use it bringing more contributions, shouldn't they?

Do load testing tools fit open-source community? Usually people who use load testing tools are not programmers and start to modify complex C++ code with MFC (what we have in case of OpenSTA) is quite difficult for them (or even impossible). Is that the problem?

We have at least 17 open source projects in that field
(http://www.opensourcetesting.org/performance.php), but I heard positive feedback from professional performance testers only for OpenSTA (that was developed as a coomercial tool by Cyrano and haven't moved far ahead since). Is it a pattern?

I wonder why start new projects when there is a good product appealing for small improvements to become a very good testing tool (some belives that it is already)? Is it just not "sexy" with C++/MFC to attract developers?

I am new to OpenSTA and open source development at all, so probably I missing something. Just want to share my thoughts while I am looking into it...
» Alexander Podelko's blog | login or register to post comments

Where OpenSTA is and where it could go...

Submitted by dansut on Wed, 13/10/2004 - 14:04.
As the most active of the OpenSTA team, I must say that I agree with most of your comments but would like to reply to some individually for clarification:
lack of analysis part
The initial idea was that a full analysis component would be provided (commercially) by Cyrano. The currently included functionality was only meant to be a taster with limited open export facilities. My opinion is that the best direction to head is better export facilities in the core toolset allowing open source or commercial analysis packages to be developed easily as seperate addons.
decay in the OpenSTA community
I covered this a bit on the OpenSTA Users Mailing List, my basic feeling is that we are stationary and the world has moved around us. The actual number of new users and downloads doesn't seem to drop... weird. It shouldn't be that difficult to get the ball rolling again - it just needs a couple more commited people.
Do load testing tools fit open-source community?
The traditional Open Source model is that a developer has a problem/need and codes to fix this. Perhaps most users of load testing tools don't have the specific development skills needed... I'm guessing those that do just hack something quick together in Perl/Python/VB... There is obviously some limiting factor going on here but I don't think it should be a stopper - I'm an open source developer and I'm still interested in making a tool like this work! I think the real answer here is possibly to create a core architecture that can easily be extended using the high level languages that load testing (QA) engineers are used to.
Commercial type tools are what performance testers like
There's no argument that OpenSTA wouldn't exist if it wasn't for its initial commercial roots, and that commercial style tools take a lot of work to maintain. This is why I am most willing to put my donated time into making OpenSTA more accessible to other developers and simpler to maintain in the future. The core functionality is what people like in that it provides a commercial type load testing environment so this doesn't need fixing - it just needs to be much easier for many more people to get involved in, which most commercial tools aren't (for obvious reasons).
Is it just not "sexy" with C++/MFC to attract developers?
IMHO developers code for 2 main reasons:
  • to make money
  • to solve their own problems
There is a third group of reasons, which are basically to satisfy intellectual problem solving urges and the joy of helping others - but these are usually enhancements to the first 2. Therefore to attract developers to OpenSTA they either need to be able to make money from it, or they need to be able to perform load tests with this type of tool (maybe they can make money providing load tests too). I don't think it's a magical formula but it is something that has barely worked up to now. I am neither making much money or doing too many load tests (although I'd like to do some of both) but I also have a long history with this project and some of its code and would like to see it reach its full potential.
I think that's enough rambling, hopefully this can mark a turning point to the OpenSTA project. I intend to make another release this year and am actively looking for ways that allow me to spend more time working on, fixing and improving the toolset. All I can say is that if you want to see OpenSTA improve faster then get involved, get your employers involved and keep spreading the word of its potential.
» login or register to post comments

Comment viewing options

Select your preferred way to display the comments and click 'Save settings' to activate your changes.