topic Re: Service load drop detection - minimum required traffic in Alerting

Service load drop detection - minimum required traffic

THellwig_VKB — Wed, 08 Jun 2022 08:11:59 GMT

Hi all,

we are alerting on a set of services for load drops. The services have between 40 and less than 10 requests per minute, no traffic at night/weekends.

Now the services with around 10 or less requests a minute did not fire a problem on the public holiday yesterday as expected, the others worked fine.

Since I can't set an 'over alerting' threshold for load drops or spikes - I guess there is a built-in one? What value is it and how to get it lowered?

kind regards,

Torsten

Re: Service load drop detection - minimum required traffic

Babar_Qayyum — Tue, 07 Jun 2022 10:24:38 GMT

Hello @THellwig_VKB

Did you try to configure the fixed thresholds for those requests?

Also, service load drops/spikes are based on the %, therefore, I think the analysis of the one-week traffic and provide us with the % dropped during the night/weekends to configure.

Just have a try.

Regards,

Babar

Re: Service load drop detection - minimum required traffic

THellwig_VKB — Wed, 08 Jun 2022 07:51:15 GMT

Hi Babar,

fixed threshold don't work because of zero traffic during night and weekend.

The %-Value is also fine - I checked those values too, compared them to the week before (went from 10/min to around 0.1/min and the threshold is 80%.

That's why my suspicion was some kind of lower boundary where the anomaly detection does not alert when the expected value is below 20 or 10 requests/min.

Re: Service load drop detection - minimum required traffic

richard_guerra — Wed, 08 Jun 2022 13:53:03 GMT

Hello, I can't help with the service load drops alert directly, but can suggest how to obtain the baseline for that service - using the Environment API v1.

GET /entity/services - Lists all available services in your environment

GET /entity/services/{meIdentifier}/baseline - Gets baseline data for the specified service

*Your token will need the API v1 permission: Access problem and event feed, metrics, and topology DataExport

I imagine that once you view the baseline Dynatrace has detected for your service, you will be able to link this back to the threshold you set for Service Load Drops.

The documentation says that there are daily and weekly patterns: https://www.dynatrace.com/support/help/how-to-use-dynatrace/problem-detection-and-analysis/problem-detection/automated-multi-dimensional-baselining