I run some basic performance tests in
C# on simple methods, but there are also legacy monolithic pieces written in
C# as well as other languages. So, when testing, I often end up working with a black box that takes a number of parameters. I can write code to get a bunch of sample data out, but making sense of it (e.g. presenting it visually is a challenge). I think that most people will not bother looking at a bunch of data or a bunch of poorly done graphics. It better tell a story at a glance.
I can draw pretty 2d graphs, surfaces, heat maps, surface heat maps, what have you. I am hoping to make the graphics so useful, that, when a somewhat trained eye looks at it, they can say "I think we have a problem there. It should not be that slow when we change from 5 to 10 securities." or "This piece is running 20% longer than last month. What feature is killing the performance now?"
I do not have much experience with presenting complex data in a good way, but I know that clever tricks do exist. Good examples are shown here: http://www.datavis.ca/gallery/timelines.php The Napoleon’s March is a great one, as it packs many-dimensional data onto a 2D piece of paper.
Can you suggest any good examples of how to visualize performance test data?