Answer by Mogalayapalli M. ·
We are deploying Dynatracce 5.0 in the pre production environment to test our products for performance. We had several issues related to pure paths. So I decided to make an exmaple app and run dyantrace against it. It is just a simple DB operation. Here are few issues I see which are very consistent with what we have observed in running against out product too.
1. Timing break down is very inconssitent. I am tracking the timings on my own in my program.
2. The pure paths are random for the same execution path
3. The execution times for the methods are very inconsistent.
Answer by Andreas G. ·
I think the PurePaths look perfectly correct to me - here are some comments
a) Your PurePaths execute very fast - between 2 and 3ms. It doesnt surprise me that there is a little time difference between each PurePath if you consider that there is other stuff running on these machines as well. You also see that some PurePaths are impacted by Garbage Collection (shown in Orange) - which happens randomly -> and that is also fine
b) Overall the PurePaths are totally identical. The only thing that is different is that you have one that also shows Auto Sensor Nodes (the ones in grey). Auto Sensors capture additional information about methods that impact performance that are not covered by our regular sensors. Depending on the Load on your system, the length and response time of your purePaths you may get more or less Auto Sensor Information. This just gives you extra value.
I hope this explains a bit that this is actually not inconsistent data. If you get the chance - try to adapt your sample app to produce longer running PurePaths with more nodes, e.g: more DB Activity.
Also - have a look at the documentation regarding Auto Sensors. They are a great additional to the PurePath nodes - especially useful in Production Environnments with many PurePaths as it captures more details on real performance hotspots