Essential terminology
| Concept | Description | Examples |
|---|---|---|
| Metrics | Numerical measurements from cloud services | CPU usage, request counts, costs |
| Logs | Event records from cloud services and apps | Errors, audit trails, access logs |
| Collection Methods | How data moves from cloud to Grafana | Scrape jobs, streams, agents |
| Data Storage | Where your observability data lives | Grafana Cloud, CloudWatch, Azure Monitor |
Collection method + storage decision = your observability architecture.
Script
Before we get into the how, let’s make sure we’re speaking the same language.
When we talk about cloud observability, we’re really talking about two types of data. First, metrics: these are the numbers. How much CPU is this EC2 instance using? How many requests hit my load balancer? What did I spend on Lambda last month?
Second, logs: these are the stories. What errors did my application throw? Who logged into that database? What API calls happened during yesterday’s outage?
Now, your cloud provider already collects all of this. AWS has CloudWatch, Azure has Azure Monitor, GCP has Cloud Monitoring.
The question is: how do you get that data into Grafana where you can actually work with it? That’s what we’re here to figure out.