Predictive questions that occur when Dynatrace tracking is connected to Splunk

[parksm]

The predicted values provided by Dynatrace are known at intervals of up to 30 minutes and 5 minutes.

However, Splunk can see future data values after 60 minutes at intervals of one minute.

Is this value calculated from the Splunk, not from the data in the Dynatrace?

Thank you in advance for your answers from the engineers.

[wolfgang_beer · Accepted solution]

Well, I would assume Splunk cannot 'see' future values but instead tries to predict future values as we do as well in our Dynatrace API.

Reason why we did restrict that function to 30min is that for many metrics the future value gets more and more uncertain in terms of its prediction precision.

Of course this very much depends on the characteristics of a metric. So if you consider a linear line showing the total memory available for a host which stays at the same value for months, its pretty easy to predict the same value into the next month 🙂

For volatile metrics such as CPU consumption this might be different.

So the short answer is, yes you can fetch detailed Dynatrace data and use any standard prediction algorithm, such as an ARIMA model to 'see' future values. A simple Python notebook can do the trick.

Best greetings,

Wolfgang