HA Alerting - Cloudwatch

Grafana Alerting
,
1

Hey there,

I’ve been thinking about using Grafana as my main alerting solution for AWS with the built-in CloudWatch data source.
The infrastructure we manage is huge and critical, thus we would like to make our Grafana highly available.
Reading through the docs of HA Grafana Alerting, I noticed that each instance evaluates the rules, and they they get deduplicated.

My question is, I’m worried that setting up Alerting in HA mode would double, triple of quadruple (depending on the number of instances) our monitoring bills, because the CloudWatch metrics would be fetched N times, N being the number of Grafana instances.

Could someone tell me if this is the case, and if it is, what could i do in order for the CW metrics to be fetched only once?

Thanks in advance

2

Generally yes, it will multiple your cost.
Grafana Cloud has some “special” alert manager, which evaluates CW metrics just once - that’s IMHO not available in the OSS.

3

@hakimhc We are wrapping up our own implementation of the HA alerting for Grafana which is load balanced and execute queries only once designed for huge and critical infrastructure like yours.

It’s available as public preview and has a Community version.