Reservation recommendations

This page explains how Compute Engine generates reservation recommendations and the parameters to configure them.

Compute Engine provides reservation recommendations to help you identify idle or underutilized on-demand reservations for the previous seven days so that you can modify or delete the reservations.

Compute Engine generates recommendations automatically based on system metrics gathered by the Cloud Monitoring service. You can configure reservations recommendations to receive more or fewer recommendations.

Pricing

There are no costs associated with using idle reservation recommendations. Using recommendations to reduce your resource usage can result in cost savings. The displayed cost savings estimate is your potential monthly savings if you adjust your VM reservation to match your actual usage. For example, if you reserved 8 VMs but consistently use only 1, you see the cost savings of downsizing your reservation to 1 VM.

Limitations

Idle and underutilized reservation recommendations are not available for the following reservations:

  • On-demand reservations that are attached to committed use discounts (CUDs)
  • On-demand reservations for virtual machine (VM) instances with TPUs

How detection of idle and underutilized reservations works

Reservation recommendations for Compute Engine are based on historical usage metrics. By default, the historical observation period is the previous 7 days. By changing the default observation period, you can customize the recommendations that you receive.

To generate recommendations, the algorithm considers reservations that accrue costs, but aren't associated with an active Compute Engine resource for the previous 7 days.

Frequency of recommendations

After a reservation is created and you haven't consumed any resources for at least 7 days, Compute Engine begins generating recommendations for it. New recommendations are generated once per day.

Customize recommendations

Compute Engine lets you customize the recommendations you receive for your project by changing the configuration used by the recommendation algorithm. In particular, by changing the default observation period, you can receive recommendations that better fit your workloads, applications, and infrastructure needs.

To learn how to modify the configuration for your project, see the following:

Choose the right configuration

This section describes the values that you can set for the configuration. Changing these values affects the recommendations that you receive.

The observation period

Set the observation period duration to calculate recommendations by modifying the value for idle_reservation_lookback_period or under_utilized_reservation_lookback_period and upload the new configuration for your project. You can set the observation period to a value between 7 days and 30 days, for example:

  • For an observation period of the previous 7 days, use "P7D".
  • For an observation period of the previous 30 days, use "P30D".

By default, the observation period is 7 days.

  • For recommendations based on short-term changes in your workload, use a shorter observation period.
  • For recommendations that are not affected by short-term fluctuations in your workload, use a longer observation period.

Similarly, set the usage threshold that triggers an underutilized reservation recommendation by modifying the value for under_utilized_reservation_utilization_threshold and upload the new configuration for your project, for example:

  • For a threshold of 80%, "0.8".

What's next