15 hours ago, Stuart Weenig said:
You can configure Dynamic Thresholds to look for deviations from the norm either positive, negative, or both. Perhaps some more information about what you've tried so we can guide you to something that will fit.
Yeah sure, to add a bit more detail -
The alert is a simple JDBC collector, runs a script against an SQL DB to get a count of what we refer to as "Flow types".
I then have a dashboard for each customer that has a line graph of all the flow types for that customer and shows the counts.
On the dashboard an example of one flow type usually ranges from a count of 500-1500 however the anomaly detection "Expected Range" (specifically the lower range) is going down to -500.
I'd expect as time goes on the expected range becomes smarter so to say based on the data it recieves?
Thanks
Jonny