kantver - Fotolia

How does the CloudWatch monitoring service track events?

We have the software and policies in place to detect events, but we need to improve our response to cloud problems. How can we use CloudWatch Events to proactively address cloud issues?

One of the fundamental benefits of public cloud services is automation -- the ability to detect and respond to particular circumstances without direct intervention from IT staff. And AWS allows admins to detect and respond to various events through Amazon CloudWatch monitoring, which includes a feature called CloudWatch Events. Through Amazon CloudWatch monitoring, IT teams can collect metrics, watch logs, and handle alarms in AWS resources, but CloudWatch Events adds the ability to automatically respond to those metrics.

CloudWatch Events keeps tabs on a spectrum of events, which most commonly includes changes to AWS resources. For example, every time an Amazon Elastic Compute Cloud (EC2) instance launches, terminates or changes state from pending to running, a detectable event occurs. Events can come from other Amazon services, and CloudTrail produces an event with every API call or AWS Management Console task. AWS users can also create custom and regularly scheduled events in applications and convey those to CloudWatch Events.

CloudWatch Events keeps tabs on a spectrum of events, which most commonly includes changes to AWS resources.

Simply detecting events through Amazon CloudWatch monitoring doesn't help much without a way to respond; CloudWatch Events can correlate events to specific responses, or targets. For example, a high-processor utilization event could trigger a selected AWS Lambda function that could launch another EC2 instance and load-balance the traffic. Similarly, events can trigger responses in Amazon Kinesis Streams for data analysis, Amazon Simple Notification Service (SNS) topics for response notifications, CloudWatch alarms and so on.

Events and responses are configured through rules. AWS users can develop detailed rules that can perform certain activities in response to events, and rules can issue multiple responses to the same event. For example, the same processor utilization event that triggers a new AWS Lambda function for more resources can also trigger an SNS notification and an alarm through the CloudWatch monitoring service.

CloudWatch Events has a few limitations. Users can only create up to 50 rules per account, and a rule can only trigger up to five events (targets). Users can invoke up to 20 events per second; additional events are throttled. Most CloudWatch API calls handle up to five requests per second. And there are additional limits in the size of event requests and list results.

Next Steps

Track EC2 instances using CloudWatch

CloudWatch logs identifies potential trouble spots

Know the limitations of CloudWatch Logs

Dig Deeper on AWS CloudWatch and application performance monitoring