Slide 4 of 7

Two approaches

Two approaches to cloud observability

Two approaches comparison: Store in Grafana Cloud vs Visualize Without Storing

ApproachBest forProduct
Store in Grafana CloudLong-term retention, unified alerting, cross-cloud visibilityCloud Provider Observability
Visualize Without StoringQuick setup, cost sensitivity, existing cloud investmentsGrafana Data Sources
Store in Self-Managed BackendData residency requirements, existing Mimir/Loki infrastructureGrafana Alloy

Which approach is right for you?

Most organizations benefit from storing data, which unlocks unified alerting, cross-cloud correlation, and long-term analysis. But if you just need quick visualizations of existing cloud data, data source connections work great too.

Note

If you’re running self-managed Grafana with your own Mimir/Loki backends, you can use Grafana Alloy to collect cloud provider data and store it in your own infrastructure. The Alloy collection methods covered in this course apply to both Grafana Cloud and self-managed scenarios.

Continue to learn more, then choose your path after the prerequisites.

Script

Here’s the big decision you need to make. You have two fundamentally different approaches to getting cloud data into Grafana, and they serve different purposes.

The first approach is to actually collect and store your data in Grafana Cloud. Your metrics go into Mimir, that’s Grafana’s scalable metrics backend, and your logs go into Loki, the logs backend. This is what we call Cloud Provider Observability.

The big advantage? All your data from all your clouds lives in one place. You query everything with PromQL and LogQL. You get unified alerting that can see across AWS, Azure, and GCP simultaneously. And you control how long data is retained.

The second approach is simpler: just connect Grafana directly to your cloud provider’s monitoring service. When you open a dashboard, Grafana queries CloudWatch or Azure Monitor in real-time. Data stays where it is. You’re just visualizing it. This is great for quick wins or when you want to keep things simple, but you lose some of the power that comes with unified storage.

Now, there’s also a third option I should mention: if you’re running self-managed Grafana with your own Mimir and Loki backends, you can use Grafana Alloy to collect cloud provider data and store it in your own infrastructure. This course focuses on Grafana Cloud, but the Alloy collection methods we’ll cover apply to both scenarios.

Think about what you’re trying to accomplish. Do you want Grafana to be your single source of truth? Go with the storage approach. Just want to visualize what’s already in CloudWatch? Skip ahead to Module 3.

/
Previous Next