[Question/Feedback]: How are the thresholds decided for AMBA alerts?

[M76-G]

Check for previous/existing GitHub issues

• I have checked for previous/existing GitHub issues

Description

Hi AMBA team

Just a quick query. I've checked a couple of the AMBA thresholds that are deployed:-

https://azure.github.io/azure-monitor-baseline-alerts/services/Sql/servers/#databases---cpu_used

vs

https://learn.microsoft.com/en-us/azure/azure-sql/database/monitoring-metrics-alerts?view=azuresql

For example, low temp db space has a threshold of 80% on AMBA, but 60% on the Monitoring Azure SQL page.

Is there a 'source of truth' somewhere that the AMBA figure is taken from, or are these 'real world' figures which have been 'tuned' for production?

Many thanks for the great work!

[JoeyBarnes]

Hi @Leeg14 - Thank you for the question! The thresholds (and other alert settings) are based on a combination of Product Group verified/recommended/documented settings and popular settings data-mined from 'real world' deployments. For the particular metric above, the 80% is coming from the popular settings (versus documented). I do appreciate you calling out the discrepancy since we do try to align these settings with any product documentation (as a priority over the popular settings)