search

Azure / azure-monitor-baseline-alerts

Azure Monitor Baseline Alerts
MIT License
123 stars 177 forks source link

[Question/Feedback]: How are the thresholds decided for AMBA alerts? #254

Open Leeg14 opened 1 week ago

Leeg14 commented 1 week ago

Check 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 commented 1 week ago

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)