Level 2 · Systems

Knowing what is happening

The site is slow. You have no metric that says so, no log you can search, and no alarm that fired. Observability is not a service you switch on afterwards; it is a design decision.

40 minCloudWatch0% completeAssumes you have done: Load balancing and scaling

Chapters

Why you should alarm on symptoms rather than causesAlarming on every CPU spike leads to paging fatigue, whereas targeting end-user symptoms ensures you only wake engineers for real outages.14 minFoundationsUnder the hood

The path

When & whyWhen to reach for CloudWatchYou need to collect, monitor, and analyze logs, metrics, and alarms across all your cloud resources in one place.