Hi
We saw a PPT in APM University about best practices and one of the best practice was to reduce the business transaction complexity.
How can we check if a BT is complex or not? is it based on number of filters we use or/and number of calculation measure we use or/and the number of splitting measures we use?
Is there some recommended numbers on filters, calculations, and splitting measure that we can use in a BT?
Thanks,
Sreerag
Answer by Sreerag M. ·
I would say we have the data from these dashboards, but to make to make more sense out of it we need to know the optimal values, like threshold for the dashboard measures.
We have skipped purepaths during peak load. We have no idea on what is causing it. Most of the case reducing purepath count in not an option, but if we have 5-6 parameters that causes the purepaths to skip then we could tweak some of those to avoid it. We need more insight into issues like these.
Based on the measure 'Skipped Events per Interval (split by Agent)' only agents in one profile is skipping purepaths. We have most of the load in this profile. Most of the purepaths are part of 7-8 BTs. Will this be impacting the server in post processing of the purepaths.
Sreerag
Answer by Herwig R. ·
Best practice is to take a look at the measure overview dashboard. It is accessible from the start center. This in combination with the server health should give you the necessary information. The 2 dashboard shows the various numbers you need to take care of. I hope this helps.
Herwig
Answer by Sreerag M. ·
We don't have too many splittings, definitely not to the range of 50K. However there is always changes in the settings and new BT is getting created. Is there a measure that can tell us that the BT complexity is impacting the dynatrace performance. Again this isn't limited to BT complexity. It there are 4-5 key things that may impact the dynatrace system performance may be dynatrace can monitor those and alert us when it start impacting the system. It is really hard to find the root cause when many changes lead to a performance hit in the system and result in skipped events and purepaths. There are many things to look at during an issue like more purepaths, less BT, BT complexity, various Buffer sizes etc. Being a monitoring tool we hope the tool can monitor its own performance better and provide inputs in fixing some of the issues.
-Sreerag
Answer by Guenter H. ·
Hi,
for sure you shouldn´t have a too big number of splittings, say, lower than 50,000. See Preventing Excessive Split Results.
G.
JANUARY 15, 3:00 PM GMT / 10:00 AM ET