Other Testing Tools
I am looking for any information, feedback etc. on the following testing tools:
Any info at all would be appreciated. Also
how would these compare to Mercury WinRunner
We will be load testing a website and would also like to do some gui regression testing.
Re: Other Testing Tools
Since you mention load and regression testing, you might want to stick with one vendor or not depending on your needs. Some people prefer to keep one tool vendor while others do not mind separate solutions. Part of this depends on budget, of course, but also on the degree of integration you want between your tools.
Optimal is good for getting some measures of your site but they are not as robust a solution (in some ways) as the offerings from Mercury, Segue, Rational, CompuWare, RSW Software, RadView, etc. I say this because they were geared to a more limited offering than the other tools in light of the testing you might be doing. They are, however, an excellent tool within their purvue (such as monitoring) and, truth be told, it has been a couple of months since I did anything with their tools. I sort of lump Benchmark Factory into this same category based on the needs you stated.
As to the others, like CompuWare, Segue, Mercury, etc. they all offer tool solutions for performance testing (of which load testing is a part) and functional/regression testing. Which one is good for you? Well, it depends on the nature of the GUI testing you are going to be doing, the objects you will be testing within that GUI, and the testing methodology you subscribe to and will implement in your use of the tools.
Each of the tools has strengths, weaknesses, and learning curves that may influence your decision. As just a simple example, SilkTest has a larger learning curve in some ways because even though you can record-and-playback with it, that is not a good way to do automation and SilkTest's back-end language, 4Test, is, in my opinion, more suited to developing scalable, robust, back-end testing engines/architectures. On the other hand, Mercury's WinRunner is very much geared to a record-and-playback paradigm because it is very procedural in nature (whereas Segue's SilkTest is object-oriented). You can develop test engines/architectures in both tools but my experience is that SilkTest is more amenable to this. I think RSW's eTest language and overall software is easier as well from a learning curve standpoint.
There is also the issue of support for the tool and the nature of your testing. If the goal is throw-away scripts (more common than you might think) then something like WinRunner might be more up your alley with its quicker turnaround on procedural tests. But also consider the objects you will be looking at. Since this is a Web site, will you have ActiveX, Java applets, or other components that you might want to test in the page. Certain tools handle these better than others.
As to performance testing solutions, consider how your performance testing is going to be conducted. Is it just load testing or will transaction, volume, or stress testing play a part? Will you be doing a robust effort to tie this into capacity planning efforts or will these just be simple tests run with the intention of determining simple metrics like response time? In any event, tools like Rational's Load Test or Segue's SilkPerformer require a higher learning curve but are much more suited for solving performance models and workload characterizations. Tools like Mercury's WinRunner, CompuWare's QALoad, or RadView's WebLoad have a lower learning curve but are not suited to some of those tasks (at least not in my opinion). You also have to consider if you need the performance tool to have hooks into application servers such as Dynamo or SilverStream. RSW and Segue offer good support for Dynamo, for example, while Mercury does not offer support for this particular application server although they do offer support for others, including SilverStream. Finally consider your performance testing needs. Are you concerned about modem simulations? If so, you are going to want a tool that can simulate them accurately (like Segue's SilkPerfomer). If you are more concerned about scheduling staged loads, you want a tool that has a good scheduling feature (like Mercury's LoadRunner 6.x).
Having said all of the above keep in mind:
1.) These are the opinions of one person.
2.) All of the solutions mentioned can be had on a trial basis to see if they fit your environment.
I highly recommend number two above because this is how you will determine if it fits in with your needs. It will also allow you to work with the scripting languages of the functional tools or the performance modeling of the performance tools and see if they fit in with how you plan to test as well as within training needs and budget.