I'm looking for practical tips to optimize spending on Kubernetes clusters. For example, I'd like to know about effective strategies like automating scale-down for developer namespaces or correctly sizing resource requests and limits. What small changes have made the biggest impact on your costs? Bonus points if you mention any automation tools you used!
4 Answers
Shutting down resources is key! Sometimes you just have to get real about it. 😂
One of the best ways we cut costs was by appropriately sizing our resource requests and limits. Pairing that with auto-scaling in production really helped minimize resource usage. We also scheduled non-production instances to scale down to zero whenever appropriate. It’s not rocket science, just getting the basics right!
We do the same in our test deployments; they scale down automatically outside business hours unless we set up exceptions.
Using spot instances has been a game changer for us. They’re a lot cheaper and make a big difference in balancing costs.
We moved our apps from ECS to EKS and avoided using NAT gateways. We also clean up dev workloads every Friday—not just to save costs over the weekend, but to keep things tidy. Deploying a single instance by default and avoiding sidecars has helped too. Just standard practices that work, nothing too flashy!
That's a solid approach! How exactly does moving from ECS to EKS make it cheaper?
I thought ECS was cheaper because it's AWS proprietary. Can you clarify?
I was thinking the same thing! Haha!