Capacity Planner overview

Private preview

This product is subject to the "Pre-GA Offerings Terms" in the General Service Terms section of the Service Specific Terms. Pre-GA products are available "as is" and might have limited support. For more information, see the launch stage descriptions.

This document gives an overview of Capacity Planner.

By using Capacity Planner, you can view the current and forecasted usage of the following Compute Engine resources in your Google Cloud project, folder, or organization:

• Virtual machine (VM) instances

• Persistent Disk volumes

• GPUs

Capacity Planner helps you plan for future capacity and quota needs by letting you do the following:

• View historical usage of your VMs, Persistent Disk volumes, and GPUs for up to two years in the past. This helps you understand your resources' usage to plan for future capacity needs.

• View forecasted usage of your VMs, Persistent Disk volumes, and GPUs for up to six months in the future. This helps you analyze and optimize your resources' usage for future growth or spikes in usage.

• View current on-demand reservations, as well as past and current future reservation requests, to plan for future growth or spikes in usage.

• Reserve resources in advance by helping you create future reservation requests. This helps you prepare for expected growth or periods when your usage doesn't follow typical patterns, such as peak scale events.

• Manage your project's quota limits by seeing how close you are to your limit and setting up automatic quota increase requests. This helps you prevent outages when reaching a quota limit, and reduce the need to submit manual quota increase requests.

Important: If you can't access Capacity Planner in the Google Cloud console, then your organization might not be eligible to access and use Capacity Planner. To request access, contact your Technical Account Manager (TAM) or the Sales team.

How Capacity Planner works

The following sections outline how Capacity Planner calculates and displays usage data and statistical forecasts for the Compute Engine resources in your project, folder, or organization.

Usage data

You can view the usage data of the VMs, Persistent Disk volumes, or GPUs in your project, folder, or organization to see their consumption patterns. The usage data in Capacity Planner is a daily percentile usage of your resources. You can apply the following usage percentile to your resources' usage data:

• P50. The P50 usage percentile is the 50th percentile usage data that separates the lower 50% of your resources' usage data from the upper 50%.

• P75. The P75 usage percentile is the 75th percentile usage data that separates the lower 75% of your resources' usage data from the upper 25%.

• P99. The P99 usage percentile is the 99th percentile usage data that separates the lower 99% of your resources' usage data from the upper 1%.

Capacity Planner samples usage data as follows:

• When you view the usage data in a project, Capacity Planner samples the usage data every five minutes and displays the data within 24 hours after usage.

• When you view the usage data in a folder or organization, Capacity Planner samples the usage data of all projects within your folder or organization every five minutes, and aggregates them to a daily usage percentile value. Then, Capacity Planner displays the data within 24 hours after usage.

Capacity Planner retains historical usage data for up to two years. However, data related to future reservation requests is only accessible for one year after the start of its reservation period.

Forecasts

You can view the forecast of the VMs, Persistent Disk volumes, or GPUs in your project, folder, or organization to help you predict your future capacity needs. The forecast in Capacity Planner is an ensemble of the following:

• A linear and superlinear model.

• If your project, folder, or organization has usage data for at least two years, then the yearly seasonality effects.

The forecast provides an estimate of future usage based on historical usage patterns for a specified period of time. You can apply the following prediction intervals to the forecast:

• P25-P75. The P25-P75 interval is a prediction interval of 50%, which means that there is a 50% probability that actual future usage will fall within the lower bound and upper bound forecast.

• P05-P95. The P05-P95 interval is a prediction interval of 90%, which means that there is a 90% probability that actual future usage will fall within the lower bound and upper bound forecast.

The statistical forecast is estimated for up to six months in the future.

What's next

• Learn how to view usage and forecast data in Capacity Planner.