This document describes how to view the actual and forecasted usage data of the following Compute Engine resources in your Google Cloud project, folder, or organization:
Virtual machine (VM) instances, including vCPUs, RAM, and Local SSD disks attached to the VMs.
Persistent Disk volumes, whether they're attached to VMs or not.
GPUs, whether they're attached to VMs or not.
Viewing the usage data of your VMs, Persistent Disk volumes, and GPUs helps you analyze and optimize their usage, as well as plan for future capacity and quota needs.
Before you begin
If you can't access Capacity Planner in the Google Cloud console, then your project, folder, or organization might not be eligible to access and use Capacity Planner. To enable your project, folder, or organization to access and use Capacity Planner, contact your Technical Account Manager or the Sales team.
If you haven't already, then see how Capacity Planner works.
Required roles
To get the permissions that you need to view the actual and forecasted usage data of your Compute Engine resources, ask your administrator to grant you the following IAM roles on your project, folder, or organization:
-
To access and view Capacity Planner:
Capacity Planner Viewer (
roles/capacityplanner.viewer
)
For more information about granting roles, see Manage access to projects, folders, and organizations.
These predefined roles contain the permissions required to view the actual and forecasted usage data of your Compute Engine resources. To see the exact permissions that are required, expand the Required permissions section:
Required permissions
The following permissions are required to view the actual and forecasted usage data of your Compute Engine resources:
-
To view the forecasted usage of your Compute Engine resources:
capacityplanner.forecasts.list
on the project, folder, or organization -
To view the historical usage of your Compute Engine resources:
capacityplanner.usageHistories.list
on the project, folder, or organization -
To summarize the historical usage of your Compute Engine resources:
capacityplanner.usageHistories.summarize
on the project, folder, or organization
You might also be able to get these permissions with custom roles or other predefined roles.
Limitations
When viewing the actual and forecasted usage data in Capacity Planner, the following limitations apply:
In the Usage and forecast by machine family graph, you can't view the quotas for your VMs and their resources in the following scenarios:
If you filter the graph by all machine families.
If you filter the graph by zone.
If you view the actual and forecasted usage data in an organization or folder.
The Usage and forecast by machine type table is available only when viewing the actual and forecasted usage data of VMs or GPUs.
View the forecast and usage of your resources
You can view the actual and forecasted usage data of the VMs, Persistent Disk volumes, and GPUs in your project, folder, or organization by doing the following:
In the Google Cloud console, go to the Capacity Planner page.
When the Capacity Planner page opens, you can then filter the UI elements to see different data visualizations for the actual and forecasted usage of the resources in your project, folder, or organization.
Optional: To change project, folder, or organization, in the project picker at the top of the page, select a project, folder, or organization.
Select the type of Compute Engine resource you want to view usage data for:
To view your VMs' usage data, click the Virtual machine tab, if it's not already selected.
To view your Persistent Disk volumes' usage data, click the Disk tab.
To view your GPUs' usage data, click the GPU tab.
The following sections outline the different UI elements on the Capacity Planner page in the Google Cloud console.
Machine family distribution for a selected location
The Machine family distribution for a selected location chart shows the distribution of a single or all machine families across the locations where you created VMs. By default, the chart shows the most used machine family in the last 30 days.
Region distribution for a selected machine family
The Region distribution for a selected machine family chart shows the distribution of a single or all machine families across all regions and zones where your VMs are located. By default, the chart shows the region with the highest usage in the last 30 days.
Usage and forecast
You can view a graph with the historical daily 50th, 75th, or 99th percentile usage and the forecasted lower bound, upper bound, and median of a selected Compute Engine resource. According to the Compute Engine resource you select, one of the following graphs appears:
The Usage and forecast by machine family graph. This graph shows the Local SSD disks, RAM, and vCPUs attached to the VMs in your project, folder, or organization, organized by machine family and location.
The Usage and forecast by disk type graph. This graph shows the Persistent Disk volumes in your project, folder, or organization that are either attached to VMs or not, organized by Persistent Disk type and location.
The Usage and forecast by GPU type graph. This graph shows the GPUs in your project, folder, or organization that are either attached to VMs or not, organized by GPU model and location.
By default, the Usage and forecast by machine family graph appears and shows the 99th percentile usage, a prediction interval of 50%, and quota for the most used machine family and region in the last 30 days.
Additionally, if you're viewing the usage and forecast data of a project's resources, you can also view the following in the graph:
The View quota toggle, which you can click to the on position to view a marker with your project's quota for the selected resource.
If your project has sufficient usage data, the Enable quota adjuster button, which you can click to enable the quota adjuster in your project.
Usage and forecast by machine type
The Usage and forecast by machine type table displays historical and predicted usage of the following Compute Engine resources:
If you're viewing the usage data of your VMs, then the graph displays the predicted usage of the VMs within a selected machine family and region.
If you're viewing the usage data of your GPUs, then the graph displays the predicted usage of the VMs that specify a selected GPU model and region.
Capacity Planner presents data at 50th, 75th, or 99th percentile, broken down by individual machine type and zone. The table includes the following columns:
Machine type: The details of each machine type in the selected machine family, including their vCPU, RAM, attached Local SSD disks (if any), and GPU type (if any).
Zone: The zone where each machine type is located.
Average of NTH percentile usage - VM count: The number of VMs using each machine type, averaged at the selected percentile.
Average of NTH percentile usage - Cores count: The number of vCPUs used per machine type, averaged at the selected percentile.
Percentage breakdown by machine type: The use of each machine type relative to other machine types in the table.
Forecast VM count - PREDICTION_INTERVAL: The forecasted number of VMs over the selected prediction interval.
Forecast Cores count - PREDICTION_INTERVAL: The forecasted number of vCPUs per machine type over the selected prediction interval.
Top projects based on historical usage: The top 20 projects with the highest historical usage of the selected machine type.*
Usage contribution by top projects: The percentage of usage of the top 20 projects for the selected machine type.*
Except for the percentage breakdown by machine type, Capacity Planner rounds all values to the nearest whole number. By default, the table shows the 99th percentile usage and a prediction interval of 50% for the machine types of the machine family with the highest usage in the most active location in the last 30 days.
* You can view this column only when viewing the usage of VMs or GPUs for a folder or organization.
What's next
Learn how to export usage and forecast data.
Learn how to enable the quota adjuster.
Learn how to reserve capacity from actual or forecasted usage.