I would like to see a stress test ability make its way into WPM, especially with the upcoming grouping of player nodes. I imagine a several player nodes in my DMZ that are used for Certification and Development all in one group separated from my Production environment. I’d like to be able to deploy a transaction from this group and have each node open X number of TCP sessions to the same web app. Maybe X is 10 each, or 100 each, or 1000. The more stress I need, the more nodes in my WPM player group. I would start out small and watch the response time and the TCP waterfall for when response times start to climb drastically. Then I would know at point “Y” things are critical mass, and at that load I would know to start looking at CPU, Memory, Network, etc.
The thing that would seem difficult would be the reporting, in my opinion you would need all the metrics from every TCP connect. Possibly even the TCP waterfall for every session. Because if you had a situation where your waterfall contained a bunch of thing but say one particular javascript item doesn’t scale well with load, but at the same time other javascripts and css files are performing normal, you would need to know that. And if you aggregate the data you may lose that. Perhaps something like a Top N for those metrics that are outliers. So I would say graphs and pie charts with aggregate data and a place to pull raw data.
In the 3 short months we have been running WPM my development staff has asked me about this 5 times.
-b
