Rationale

We use CloudWatch for monitoring our entire AWS infrastructure. We can monitor our applications, react to performance changes within them, optimize resource utilization, and get a unified view of operational health. The main reasons why we chose it over other alternatives are the following:

• It is a core AWS service. Once we start creating infrastructure, CloudWatch begins to monitor it.
• It integrates seamlessly with most AWS services. Some examples are EC2, S3, and DynamoDB.
• It complies with several certifications from ISO and CSA. Many of these certifications are focused on granting that the entity follows best practices regarding secure cloud-based environments and information security.
• It supports custom dashboards for visualizing metrics using diagrams like bars, pies, numbers, among others. Other customizations such as timespans and resource metrics as axes are also available.
• It supports alarms using AWS SNS, allowing email notifications to be triggered when resource metric conditions are not met or anomalies are detected.
• Resources can be written as code using Terraform.

Alternatives

Note: GCP Cloud Monitoring and Azure Monitor are alternatives that did not exist at the time we migrated to the cloud. A review of each of them is pending.

Usage

We use CloudWatch for monitoring

• EC2 instance performance;
• EBS disk usage and performance;
• S3 bucket size and object number;
• ELB load balancer performance;
• Redshift database usage and performance;
• DynamoDB tables usage and performance;
• SQS sent, delayed, received and deleted messages;
• ECS cluster resource reservation and utilization, and
• Lambda invocations, errors, duration, among others.

We do not use CloudWatch for

• Synthetic monitoring (we use Checkly instead);
• ServiceLens (it only supports Lambda functions, API Gateway, and Java-based applications);
• Contributor Insights (we use Cloudflare instead);
• Container Insights (we use New Relic; pending review);
• Lambda Insights (we currently use Lambda for a few non-critical tasks);
• CloudWatch agent (it could increase visibility for EC2 machines; pending review);
• CloudWatch Application Insights (it only supports Java-based applications), or
• writing our alarms as code using Terraform (pending to be done).