06 Apr 2021 11:26 AM - last edited on 09 Apr 2021 03:52 AM by Karolina_Linda
What would be the advantage of Davis AI over the error budget burn rates to notify? Does this alert system take into account the attributes of precision, recall, detection time, and reset time?
Solved! Go to Solution.
I'm not sure I 100% understand the question you are asking. Could you please be more clear?
Are you trying to understand what the advantage of the Davis alerting system is over alerting off of SLO's?
SLO's do not take into consideration many sorts of metrics. For example if you had a garbage collection issue that you wanted to catch before it impacted your end users that would not be possible with SLO based alerting.
Hello Dallas, thank you for answering me.
I understand what you are saying, but my questions are more oriented to the effectiveness of the Davis alert system to defend my SLOs so that I can react in time when I am consuming my error budget very quickly, and at the end of the period I do not arrive with this exhausted budget. So I hope that Davis's detection time is correct and that it only notify me when there are significant events that put my SLO at risk and stop warning me when the issue has been resolved.
So the question would be: Would Davis help me protect my SLOs or would I have to create other types of alerts in Dynatrace?
Yes, the out of the box alerts should be helpful in protecting your SLO's. Of course they are only going be be a warning system before your system degrades.
That being said there is going to be a constant effort involved in fine tuning the alerting system to fit your environment. I would recommend learning about the out of the box alerts before configuring custom ones. see the document here:
Let me know if you have any more questions.
Hi @Dallas, I found this post in your product ideas and roadmap section. Apparently, you are working on these features
Drill-down from SLOs to related DAVIS AI detected problems - even before SLOs are violated
Custom calculations with math with multiple metrics to define your custom 0-100% normalisation
Alerting on SLO violations (as alerting will become available for any calculated metric)
Alerting on Error Budget Burn Rate (as alerting will become available for any calculated metric)
Use SLOs in automated quality gates within your software delivery (Shift-left of SLO evaluation)
Is there any other information about it?