---
title: "Metrics key takeaways | Grafana Labs"
description: "Key takeaways for collecting cloud metrics."
---

> For a curated documentation index, see [llms.txt](/llms.txt). For the complete documentation index, see [llms-full.txt](/llms-full.txt).

## Key takeaways

- Cloud providers expose metrics across compute, containers, databases, storage, and networking, and Grafana can collect all of them.
- You have three collection methods:
  
  - **Scrape jobs**: the simplest start, configured in the Grafana Cloud UI with no infrastructure to deploy. Available for AWS and Azure.
  - **Metric streams**: AWS pushes metrics for low-latency delivery, which suits production alerting.
  - **Grafana Alloy**: flexible and works across clouds, and it’s the collection path for GCP.
- Choose a collection method based on latency, scale, and how much you want to run yourself: scrape jobs for a quick start, metric streams for low-latency AWS, and Alloy for multi-cloud or GCP.