Hi All, Just wanted to share some good achievement by me in my current project. Successfully reduced almost 45 % cost savings on Azure Kubernetes, by figuring out the right instance series should be used, weekend power off management, out of business hour powering off and also used spot instance where the downtime is acceptable mostly in maintenance app in dev stage.

... by figuring out the right instance series should be used

In AWS I've been using Karpenter and works really good as well. For cloud-agnostic solutions I've been using KEDA with Cluster API What kind of tools have you used to approach your solutions ?

I have not used any specific tools here, however I saw there are few node pools created with high configuration such as F series, which was created while ago for development and testing, from the monitoring section which AKS gives to see the cpu, ram utilisation etc from their I could able to see that the application is not utilizing resource with full capacity like not even 30 to 40 percent in business days/working hours. Hence I concluded that the instance series should be downgraded (chooses actually D series where we can have a balance of cpu and ram). And also used an E series instance where one single node can handle a load of two d series node. In this way I was able to reduce the cost

Nice research ! 🚀