Compute Engine rate quotas and system limits

This document lists the Compute Engine rate quotas and system limits that specify the number of requests you can make to Compute Engine API methods.

Quotas restrict how much of a Cloud de Confiance by S3NS resource your project can use. If you want to change your quota value to accommodate more resource use, you can request a quota adjustment. System limits are fixed constraints which cannot be increased or decreased. For more information, see the Cloud Quotas overview.

Rate quotas and system limits

Rate quotas (also known as API quotas) define the number of requests that can be made to the Compute Engine API. These quotas apply on a per-project basis. Each quota applies to a group of one or more Compute Engine API methods. When you use the gcloud CLI or the Cloud de Confiance console, you are also making requests to the API and these requests count towards your rate quota. If you use service accounts to access the API, that also counts towards your rate quota.

Google enforces rate quotas at a per-minute interval for each group. If your project reaches the maximum number of API requests within 60 seconds, you must wait for that rate quota to refill before making more requests in that group. For example, if a project reaches a quota limit at 10:00:15, the quota refills at the start of the next synchronized interval, such as 10:01:00, rather than refilling immediately. If your project exceeds a rate quota, you receive a 403 error with the reason rateLimitExceeded. To resolve this error, wait a minute then try your request again—the quota should be refilled at the start of the next interval.

The number of requests that you can make to the Compute Engine API is defined by API quotas as described in the following tables. Each group is counted separately, so you can achieve the maximum quota in each group simultaneously.

Rate quotas and system limits for global methods

The following sections list the quotas that define the number of requests you can make to the global Compute Engine API methods.

Simplified API quotas and system limits for global methods

To improve the discoverability and manageability of Compute Engine API quota, Cloud de Confiance has reduced the number of quota metrics by consolidating metrics into fewer quotas. These simplified quota metrics also offer higher limits for each quota.

To view a complete list of global methods and the quota metrics that track the cost of each method, see Compute Engine API quota metrics reference.

The following table lists the simplified API quotas and metrics for Compute Engine API:

Quota and description Metric name
Read requests per minute (GlobalReadsPerMinutePerProject)
Quota for global get and list methods. 		compute.googleapis.com/global_reads
Write requests per minute (GlobalWritesPerMinutePerProject)
Quota for global write methods that are not included under other quota metrics. 		compute.googleapis.com/global_writes
Cache invalidation requests per minute (GlobalCacheInvalidationRequestsPerMinutePerProject)
Quota for global UrlMapsService.InvalidateCache methods. 		compute.googleapis.com/global_cache_invalidation_requests

The following table lists the system limits and metrics for Compute Engine API:

System limit and description Metric name
List usable requests per minute (GlobalListUsablePerMinutePerProject)
System limit for global ListUsable methods. 		compute.googleapis.com/list_usable_requests
Filtered list cost overhead (ListRequestsFilterCostOverheadPerMinutePerProject)

System limit for *.list and *.aggregatedList global methods with filters.

Cloud de Confiance charges quota usage for this metric in addition to the quotas against the compute.googleapis.com/global_reads metric. You incur quota charges if there are more than 10k resources filtered out of the list requests. Compute Engine API rejects the list requests if you exceed this quota limit.

 compute.googleapis.com/filtered_list_cost_overhead
Rate per project: 750k resources filtered out of the list requests per region per minute. As the quota is charged for every 10k resources, you see the limit as 75 (750k/10k) when you query for this quota limit on the Cloud de Confiance console.
Requests per minute (GlobalRequestsPerMinutePerProject)

System limit for all global read and write requests.

 compute.googleapis.com/global_requests

Legacy API quotas and system limits for global methods

The following table lists all Compute Engine API quotas for global methods.

Quota and description Metric name
Queries per minute (defaultPerMinutePerProject)
Quota for global list and mutation methods that are not included under other quota metrics.

The following get methods also use this default metric:

  • networkFirewallPolicies.get
  • projects.getXpnHost
  • projects.getXpnResources
 compute.googleapis.com/default
Read requests (legacy) (ReadRequestsPerMinutePerProject)
Quota for global *.get methods.		 compute.googleapis.com/read_requests
List requests (ListRequestsPerMinutePerProject)
Quota for global *.list methods.		 compute.googleapis.com/list_requests
Operation read requests (OperationReadRequestsPerMinutePerProject)
Quota for globalOperations.get method.		 compute.googleapis.com/operation_read_requests
Global resource mutation requests (GlobalResourceWriteRequestsPerMinutePerProject)
Quota for images.delete, images.deprecate, images.insert, images.setLabels, snapshots.delete, snapshots.insert, snapshots.setLabels, machineImages.insert, and machineImages.delete methods.		 compute.googleapis.com/global_resource_write_requests
Heavy-weight mutation requests (HeavyWeightWriteRequestsPerMinutePerProject)
Quota for patch, delete, and insert methods for the interconnects resources.		 compute.googleapis.com/heavy_weight_write_requests
Heavy-weight read requests (HeavyWeightReadRequestsPerMinutePerProject)
Quota for *.aggregatedList methods.		 compute.googleapis.com/heavy_weight_read_requests

The following quotas apply to global APIs with per method quotas:

Quota and description Metric name Default limit
License insert requests
Quotas for licenses.insert method.		 compute.googleapis.com/license_insert_requests
  • Quota per project (LicenseInsertRequestsPerMinutePerProject): 2.5 requests/second (150 requests/minute)
  • Quota per day per project (LicenseInsertRequestsPerDayPerProject): 30 requests/day
Project set common instance metadata requests
Quota for projects.setCommonInstanceMetadata method.		 compute.googleapis.com/project_set_common_instance_metadata_requests Quota per project (ProjectSetCommonInstanceMetadataRequestsPerMinutePerProject): 36 requests/minute

Rate quotas and system limits for regional and zonal methods

The following sections list all quotas that apply to methods that use regional metrics.

Simplified API quotas and limits for regional and zonal methods

The following table lists the simplified API quota metrics for Compute Engine API regional and zonal methods: To view a complete list of regional and zonal methods, and the quota metrics that track the usage of each method, see Compute Engine API quota metrics reference.

Quota and description Metric name
Read requests per minute per region (ReadRequestsPerMinutePerProjectPerRegion)
Quota for regional and zonal `get` and `list` methods.		 compute.googleapis.com/reads_per_region
Write requests per region (WritesPerMinutePerProjectPerRegion)
Quota for regional and zonal write methods that are not included under other quota metrics. 		compute.googleapis.com/writes_per_region

The following table lists the system limits for Compute Engine API regional and zonal methods:

System limit and description Metric name
List usable requests per region (ListUsablePerMinutePerProjectPerRegion)
System limit for regional and zonal ListUsable methods. 		compute.googleapis.com/list_usable_requests_per_region
Filtered list cost overhead per region

System limit for *.list and *.aggregatedList regional and zonal methods with filters.

Cloud de Confiance charges quota usage for this metric in addition to the quotas against the compute.googleapis.com/reads_per_region metric. You incur quota charges if there are more than 10k resources filtered out of the list requests. Compute Engine API rejects the list requests if you exceed this limit.

 compute.googleapis.com/filtered_list_cost_overhead_per_region
Rate per project (ListRequestsFilterCostOverheadPerMinutePerProjectPerRegion): 750k resources filtered out of the list requests per region per minute. As the quota is charged for every 10k resources, you see the limit as 75 (750k/10k) when you query for this quota limit on the Cloud de Confiance console.
Requests per minute per region (RequestsPerMinutePerProjectPerRegion)

System limit for all regional read and write requests.

 compute.googleapis.com/requests_per_region

Legacy API quotas and system limits for regional methods

The following table lists all Compute Engine API quotas for regional and zonal methods.

Quota and description Metric name
Queries (QueriesPerMinutePerRegion)
Quota for regional and zonal methods that create, modify, or delete Compute Engine resources. For example, instances.insert, disks.update, and instances.delete methods.

The following get, list and patch methods also use this default_per_region metric:

  • projects.listXpnHosts
  • instances.getScreenshot
  • instances.getGuestAttributes
  • instances.getShieldedInstanceIdentity
  • instances.getEffectiveFirewalls
  • projects.listXpnHosts
  • instanceGroupManagers.listManagedInstances
  • instanceGroupManagers.listErrors
  • instanceGroupManagers.listPerInstanceConfigs
  • regionInstanceGroupManagers.listManagedInstances
  • regionInstanceGroupManagers.listErrors
  • regionInstanceGroupManagers.listPerInstanceConfigs
  • resourcePolicies.patch
 compute.googleapis.com/default_per_region
Read requests (legacy) (ReadRequestsPerMinutePerRegion)
Quota for regional and zonal get methods such as autoscalers.get, disks.get, instances.get, and machineTypes.get.		 compute.googleapis.com/read_requests_per_region
List requests (ListRequestsPerMinutePerRegion)
Quota for regional and zonal list methods such as autoscalers.list, disks.list, instances.list, and machineTypes.list.		 compute.googleapis.com/list_requests_per_region
Operation read requests (OperationReadRequestsPerMinutePerRegion)
Quota for regionOperations.get and zoneOperations.get methods.		 compute.googleapis.com/operation_read_requests_per_region
Global resource mutation requests per region (GlobalResourceWriteRequestsPerMinutePerProjectPerRegion)
Quota for disks.createSnapshot, and regionDisks.createSnapshot methods.		 compute.googleapis.com/global_resource_write_requests_per_region
Instance get serial port output requests (GetSerialPortOutputRequestsPerMinutePerProjectPerRegion)
Quota for instances.getSerialPortOutput method.		 compute.googleapis.com/get_serial_port_output_requests_per_region
Heavy weight read requests (HeavyWeightReadRequestsPerMinutePerRegion)
Quota for regionOperations.wait, zoneOperations.wait, and regionNetworkFirewallPolicies.getEffectiveFirewalls methods.		 compute.googleapis.com/heavy_weight_read_requests_per_region
Heavy weight mutation requests per region HeavyWeightWriteRequestsPerMinutePerProjectPerRegion
Quota for patch, delete, and insert methods for interconnectAttachments resource and for the networks.updatePeering method.		 compute.googleapis.com/heavy_weight_write_requests_per_region
Instance simulate maintenance event requests (SimulateMaintenanceEventRequestsPerMinutePerProjectPerRegion)
Quota for instances.simulateMaintenanceEvent method.		 compute.googleapis.com/simulate_maintenance_event_requests_per_region
Instance list referrer requests InstanceListReferrersRequestsPerMinutePerProjectPerRegion
Quota for instances.listReferrers method.		 compute.googleapis.com/instance_list_referrers_requests_per_region
Network endpoint write requests NetworkEndpointWriteRequestsPerMinutePerProjectPerRegion
Quota for networkEndpointGroups.attachNetworkEndpoints and networkEndpointGroups.detachNetworkEndpoints methods.		 compute.googleapis.com/network_endpoint_write_requests_per_region
Network endpoint list requests (NetworkEndpointListRequestsPerMinutePerProjectPerRegion)
Quota for networkEndpointGroups.listNetworkEndpoints method.		 compute.googleapis.com/network_endpoint_list_requests_per_region
Regional network endpoint write requests (RegionalNetworkEndpointWriteRequestsPerMinutePerProjectPerRegion)
Quota for regionNetworkEndpointGroups.attachNetworkEndpoints and regionNetworkEndpointGroups.detachNetworkEndpoints methods.		 compute.googleapis.com/regional_network_endpoint_write_requests_per_region
Regional network endpoint list requests (RegionalNetworkEndpointListRequestsPerMinutePerProjectPerRegion)
Quota for regionNetworkEndpointGroups.listNetworkEndpoints method.		 compute.googleapis.com/regional_network_endpoint_list_requests_per_region

The following table lists the legacy system limits for Compute Engine API regional and zonal methods:

System limit and description Metric name
Filtered list cost overhead

System limit for *.list and *.aggregatedList methods with filters.

Cloud de Confiance charges quota usage for this metric in addition to the quotas against the compute.googleapis.com/list_requests_per_region and compute.googleapis.com/heavy_weight_read_requests_per_region metrics. You incur quota charges if there are more than 10k resources filtered out of the list requests. Compute Engine API rejects the list requests if you exceed this limit.

 compute.googleapis.com/filtered_list_cost_overhead_per_region
Rate per project (ListRequestsFilterCostOverheadPerMinutePerProjectPerRegion): 750k resources filtered out of the list requests per region per minute. As the quota is charged for every 10k resources, you see the limit as 75 (750k/10k) when you query for this quota limit on the Cloud de Confiance console.

Request an increase in rate quotas

If you need a higher quota for making API requests, you can request an increase in the API quota from the Cloud de Confiance console. For instructions, see Request a quota adjustment.

Best practices

Follow the Compute Engine API best practices for preserving rate quotas to mitigate the effects of rate quotas.

What's next