You only can control, what you measure! So CloudWatch is a possibility for monitoring your resources and applications running on AWS by collecting and tracking standard and custom metrics, like log files. If the defined alarm state is achieved, CloudWatch Alarm sends via SNS e.g. a email notification to the ops team, automatically EC2 starts with EC2 Auto Scaling, based on the rule set you defined before or if a server life cycle event „server failed“ a programmatic response could be triggered to mitigate, e.g. restart the server.
This three graphs are a good example to show case, that scaling matters! Always run the required capacity and scale up or down, if needed. The advantage is, you don’t pay for unused capacity and you don’t run in peek times out of capacity. This process is covered by EC2 Auto Scaling, to automated the provided capacity and CloudWatch is an additional monitoring Service to centralize the monitoring metrics.