How is cost calculated?
How is machine cost calculated?
Cloud providers charge per machine based on size, location, and preemptibility. Size is determined by machine type. Location is determined by region. Preemptibility is determined by whether you are using spot instances or on-demand instances (GCE, EC2). Pricing from the API is typically per-hour, but Flightcrew breaks the cost down per-minute.
The overall cost you see on Flightcrew's home page is based on what it takes to run your infrastructure: the cost of machines, the cost to run a maintained version of Kubernetes (GKE, EKS). Factoring in the cost of disk and network is coming soon.
Flightcrew does not compute discounts (e.g. 3-year contracts, etc.). If you would like us to consider discounts, please contact us on Slack or ask for support.
How is Kubernetes workload cost calculated?
| Type | Methodology |
|---|---|
| Node | The cost of the machine that is registered as the node. |
| Cluster | The cost to run a maintained version of Kubernetes and the cost of all nodes in the cluster. |
| Container | The percentage of CPU and memory resources, priced differently, that are being used or requested on the node by the container. The max of usage or requested resource is considered since only that container has access to the resource, based on how Kubernetes handles requests. |
| Pod | The cost of all containers in the pod. |
| Controllers | The cost of all pods managed by the controller. For example, a deployment with 3 replicas will include the cost of its 3 pods. |