# GCP VM Cost Analysis

>節錄自：qwiklabs `GSP767` **Exploring Cost-optimization for GKE Virtual Machines**

---

## Cost Analysis

You're now running the same workload which required three `e2-medium` machines on one `e2-standard-2` machine.

Take a look at the hourly cost for having an e2 standard and shared core machine types up:



The cost of three `e2-medium` machines would be about `$0.1` an hour while one `e2-standard-2` is listed at about `$0.067` an hour.

Saving `$.04` an hour may seem small, but this cost can add up over the lifetime of a running application. It would be even more noticeable at a larger scale too. Because the `e2-standard-2` machine can pack your workload more efficiently and there's less unused space, the cost of scaling up would grow less quickly.

This is interesting because `E2-medium` is a shared cored machine type which is designed to be cost effective for small, non resource intensive applications. But, for the `Hello-App`'s current workload, you see that using a node pool with a larger machine type ends up being a more cost effective strategy.

In the **Cloud Console**, you should still be on the **Nodes** tab of your **hello-demo** cluster. Refresh this tab and examine the `CPU Requested` and `CPU Allocatable` fields for your larger-pool node.

Notice there's room for further optimization. The new node could fit another replica of your workload without needing to provision another node. Or again, you could potentially choose a custom sized machine type that fits the CPU and memory needs of the application saving even more resources.

It should be noted that these prices will vary depending on the location of your cluster. The next part of this lab will deal with selecting the best region and managing a regional cluster.

###### tags: `GCP`