Open srpomeroy opened 2 years ago
There are likely other best practices we should consider when making node size recommendations.
A big +1 here. Other things to keep in mind are storage limitations, IOPS, etc. We should also not be recommending just 1-2 nodes on a production cluster.
This issue has been marked as stale because it has been open for 360 days with no activity. Please remove the stale label or comment or this issue will be closed in 5 days.
This is still front of mind for many prospects and customers.
What problem are you trying to solve? I would like to resize my cluster nodes for more efficient utilization using Kubecosts "Rightsize your cluster nodes" savings recommendations.
Describe the solution you'd like The recommended instance sizes and quantities should abide by Kubernetes best practices, or in the case of a cloud-based environment, the cloud providers' best practices and limitations.
Kubernetes: Considerations for large clusters
Azure: Azure Kubernetes Service service limits
AWS: Amazon EKS - Elastic Network Interface (ENI) max pods ENI max pods by instance
Google: GKE max pods per node
Describe alternatives you've considered Manually calculating instance quantities and sizes.
How would users interact with this feature? I can envision a few different ways. Looks like 110 pods per node is the most common maximum. Kubecost could set that as the default quantity and then make it adjustable via the advanced settings in the UI.
A future version may take provider maximums into consideration based on provider details, such as AWS EKS nodes using ENI.
gz#2241
(related to Zendesk ticket kubecost/cost-analyzer-helm-chart#2241)
┆Issue is synchronized with this Jira Task by Unito