Is Using CloudWatch for Monitoring EKS a Smart Move?

From what I've seen, CloudWatch logs can incur steep costs. Prometheus and Grafana are widely regarded as superior tools for Kubernetes monitoring with richer ecosystems. Plus, it could be beneficial in the long run to get familiar with them since many companies use them. But if you're looking to get metrics from AWS, CloudWatch might still be necessary to kick things off.

CloudWatch can get pricey really quickly—especially with custom metrics. Each custom metric costs around 30 cents and that can multiply depending on how you’re using it, possibly leading to thousands of dollars in expenses if you’re not careful. I’d suggest leaning towards Grafana and Prometheus instead if you're concerned about costs; they offer better control and a more rewarding setup process.

Honestly, running four clusters for just 15-20 pods each seems excessive. It might be worth considering consolidating those clusters to cut down on costs. If you stick with CloudWatch, definitely think about using Container Insights to get pod-level metrics. Just be prepared for the potential increase in metric volume and payments.

For a setup like yours, CloudWatch is likely easier to manage, but the bills will be something to watch out for. Each enabled metric in CloudWatch can add up quickly, especially if you’re tracking usage per pod and node. Managed services like Amazon Managed Prometheus and Grafana might be worth a look since they provide out-of-the-box monitoring for EKS at a potentially reduced cost.