gcloud alpha compute future-reservations create Stay organized with collections Save and categorize content based on your preferences.
- NAME
- gcloud alpha compute future-reservations create - create a Compute Engine future reservation
- SYNOPSIS
gcloud alpha compute future-reservations createFUTURE_RESERVATION--[no-]auto-delete-auto-created-reservations(--source-instance-template=SOURCE_INSTANCE_TEMPLATE| [--machine-type=MACHINE_TYPE:--accelerator=[count=COUNT],[type=TYPE]--local-ssd=[count=COUNT],[interface=INTERFACE],[size=SIZE]--maintenance-freeze-duration=MAINTENANCE_FREEZE_DURATION--maintenance-interval=MAINTENANCE_INTERVAL--min-cpu-platform=MIN_CPU_PLATFORM] | [--tpu-version=TPU_VERSION:--chip-count=CHIP_COUNT--workload-type=WORKLOAD_TYPE])(--start-time=START_TIME(--duration=DURATION|--end-time=END_TIME))[--deployment-type=DEPLOYMENT_TYPE][--description=DESCRIPTION][--name-prefix=NAME_PREFIX][--planning-status=PLANNING_STATUS][--[no-]require-specific-reservation][--reservation-mode=RESERVATION_MODE][--reservation-name=RESERVATION_NAME][--scheduling-type=SCHEDULING_TYPE][--total-count=TOTAL_COUNT][--zone=ZONE][--auto-created-reservations-delete-time=AUTO_CREATED_RESERVATIONS_DELETE_TIME|--auto-created-reservations-duration=AUTO_CREATED_RESERVATIONS_DURATION][--commitment-name=COMMITMENT_NAME--commitment-plan=COMMITMENT_PLAN--previous-commitment-terms=PREVIOUS_COMMITMENT_TERMS][--share-setting=SHARE_SETTING--share-with=PROJECT,[PROJECT,…]][GCLOUD_WIDE_FLAG …]
- DESCRIPTION
(ALPHA)Create a Compute Engine future reservation.- EXAMPLES
- To create a Compute Engine future reservation by specifying VM properties usingan instance template, run:
gcloudalphacomputefuture-reservationscreatemy-future-reservation--total-count=1000--start-time=2022-11-10--end-time=2022-12-10--name-prefix=prefix-reservation--source-instance-template=example-instance-template--zone=fake-zoneTo create a Compute Engine future reservation by directly specifying VMproperties, run:
gcloudalphacomputefuture-reservationscreatemy-future-reservation--total-count=1000--start-time=2022-11-10--end-time=2022-12-10--name-prefix=prefix-reservation--machine-type=custom-8-10240--min-cpu-platform="Intel Haswell"--accelerator=count=2,type=nvidia-tesla-v100--local-ssd=size=375,interface=scsi - POSITIONAL ARGUMENTS
FUTURE_RESERVATION- Name of the future reservation to create.
- REQUIRED FLAGS
--[no-]auto-delete-auto-created-reservations- If specified, the auto-created reservations for a future reservation are deletedat the end time (default) or at a specified delete time. Use
--auto-delete-auto-created-reservationsto enable and--no-auto-delete-auto-created-reservationsto disable. - To create a future reservation request, specify the properties of the resourcesthat you want to reserve and when you want to start using them. After therequest is approved, Compute Engine automatically creates reservations for yourrequested resources at your specified start time.
Exactly one of these must be specified:
--source-instance-template=SOURCE_INSTANCE_TEMPLATE- The url of the instance template that will be used to populate the fields of thereservation. Instance properties can not be defined in addition to sourceinstance template.
- Define individual instance properties for the specific SKU reservation.
--machine-type=MACHINE_TYPE- The type of machine (name only) that has a fixed number of vCPUs and a fixedamount of memory. You can also specify a custom machine type by using thepattern
custom-number_of_CPUs-amount_of_memory-for example,custom-32-29440.This flag argument must be specified if any of the other arguments in this groupare specified.
--accelerator=[count=COUNT],[type=TYPE]- Manage the configuration of the type and number of accelerator cards attached.
count- The number of accelerators to attach to each instance in the reservation.
type- The specific type (e.g.
nvidia-tesla-k80for nVidia Tesla K80) ofaccelerator to attach to instances in the reservation. Usegcloud computeaccelerator-types listto learn about all available acceleratortypes.
--local-ssd=[count=COUNT],[interface=INTERFACE],[size=SIZE]- Manage the size and the interface of local SSD to use. Seehttps://cloud.google.com/compute/docs/disks/local-ssdfor more information.
interface- The kind of disk interface exposed to the VM for this SSD. Valid values are
scsiandnvme. SCSI is the default and is supported bymore guest operating systems. NVME may provide higher performance. size- The size of the local SSD in base-2 GB.
count- The number of local SSD to use per VM. If you don't specify this argument, thenthe default value is 1.
--maintenance-freeze-duration=MAINTENANCE_FREEZE_DURATION- Specifies the amount of hours after instance creation where the instance won'tbe scheduled for maintenance, e.g.
4h,2d6h. See $gcloud topic datetimes forinformation on duration formats. --maintenance-interval=MAINTENANCE_INTERVAL- Specifies how infrastructure upgrades should be applied to the VM.
MAINTENANCE_INTERVALmust be one of:PERIODIC- VMs receive infrastructure and hypervisor updates on a periodic basis,minimizing the number of maintenance operations (live migrations orterminations) on an individual VM. Security updates will still be applied assoon as they are available.
RECURRENT- VMs receive infrastructure and hypervisor updates on a periodic basis,minimizing the number of maintenance operations (live migrations orterminations) on an individual VM. This may mean a VM will take longer toreceive an update than if it was configured for AS_NEEDED. Security updates willstill be applied as soonas they are available. RECURRENT is used for GEN3 andSliceof Hardware VMs.
--min-cpu-platform=MIN_CPU_PLATFORM- Optional minimum CPU platform of the reservation to create.
- You must define the version and number of TPUs to reserve.
--tpu-version=TPU_VERSION- The version of Cloud TPU to reserve.
TPU_VERSIONmust beone of:TPU7X- Cloud TPU v7x
V5E- Cloud TPU v5e Lite
V5P- Cloud TPU v5p
V6E- Cloud TPU v6e
--chip-count=CHIP_COUNT- The number of chips to reserve.
--workload-type=WORKLOAD_TYPE- The workload type of the TPU reservation.
WORKLOAD_TYPEmust be one of:BATCH- Reserved resources will be optimized for BATCH workloads, such as ML training.
SERVING- Reserved resources will be optimized for SERVING workloads, such as ML inference
- Manage the time specific properties for requesting future capacity
This must be specified.
--start-time=START_TIME- Start time of the Future Reservation. The start time must be an RFC3339 validstring formatted by date, time, and timezone or "YYYY-MM-DDTHH:MM:SSZ"; whereYYYY = year, MM = month, DD = day, HH = hours, MM = minutes, SS = seconds, and Z= timezone (i.e. 2021-11-20T07:00:00Z).
This flag argument must be specified if any of the other arguments in this groupare specified.
- Exactly one of these must be specified:
--duration=DURATION- Alternate way of specifying time in the number of seconds to terminate capacityrequest relative to the start time of a request.
--end-time=END_TIME- End time of the Future Reservation. The end time must be an RFC3339 valid stringformatted by date, time, and timezone or "YYYY-MM-DDTHH:MM:SSZ"; where YYYY =year, MM = month, DD = day, HH = hours, MM = minutes, SS = seconds, and Z =timezone (i.e. 2021-11-20T07:00:00Z).
- OPTIONAL FLAGS
--deployment-type=DEPLOYMENT_TYPE- The deployment type for the reserved capacity.
DEPLOYMENT_TYPEmust be one of:DENSE- DENSE mode is for densely deployed reservation blocks.
FLEXIBLE- FLEXIBLE mode is for highly flexible, logical reservation blocks.
--description=DESCRIPTION- An optional description of the future reservation to create.
--name-prefix=NAME_PREFIX- A name prefix for the auto-created reservations when capacity is delivered atthe start time. Each auto-created reservation name starts with the name prefix.
--planning-status=PLANNING_STATUS- The planning status of the future reservation. The default value is DRAFT. Whilein DRAFT, any changes to the future reservation's properties will be allowed. Ifset to SUBMITTED, the future reservation will submit and its procurementStatuswill change to PENDING_APPROVAL. Once the future reservation is pendingapproval, changes to the future reservation's properties will not be allowed.
PLANNING_STATUSmust be one of:DRAFT- Default planning status value.
SUBMITTED- Planning status value to immediately submit the future reservation.
--[no-]require-specific-reservation- Indicate whether the auto-created reservations can be consumed by VMs with "anyreservation" defined. If enabled, then only VMs that target the auto-createdreservation by name using
--reservation-affinity=specificcanconsume from this reservation. Auto-created reservations delivered with thisflag enabled will inherit the name of the future reservation. Use--require-specific-reservationto enable and--no-require-specific-reservationto disable. --reservation-mode=RESERVATION_MODE- The mode of the reservation.
RESERVATION_MODEmust beone of:CALENDAR- This indicates to create a future reservation in calendar mode, which is idealfor reserving GPU VMs. The auto-created reservations for the future reservationare automatically deleted at the end of the reservation period.
DEFAULT- This indicates to create a standard future reservation. If you want toautomatically delete the auto-created reservations, then you must use the--auto-delete-auto-created-reservations flag.
--reservation-name=RESERVATION_NAME- Name of reservations where the capacity is provisioned at the time of deliveryof future reservations. If the reservation with the given name does not existalready, it is created automatically at the time of Approval with INACTIVE statetill specified start-time. Either provide the reservation_name or a name_prefix.
--scheduling-type=SCHEDULING_TYPE- Maintenance for the reserved capacity.
SCHEDULING_TYPEmust be one of:GROUPED- In GROUPED mode, maintenance on all reserved instances is synchronized.
INDEPENDENT- In INDEPENDENT mode, maintenance is not synchronized for this reservation, andeach instance has its own maintenance window.
--total-count=TOTAL_COUNT- The total number of instances for which capacity assurance is requested at afuture time period.
--zone=ZONE- Zone of the future reservation to create. If not specified and the
property isn't set, you mightbe prompted to select a zone (interactive mode only).compute/zoneTo avoid prompting when this flag is omitted, you can set the
property:compute/zonegcloudconfigsetcompute/zoneZONEA list of zones can be fetched by running:
gcloudcomputezoneslistTo unset the property, run:
gcloudconfigunsetcompute/zoneAlternatively, the zone can be stored in the environment variable
.CLOUDSDK_COMPUTE_ZONE - Manage the auto-delete time properties.
At most one of these can be specified:
--auto-created-reservations-delete-time=AUTO_CREATED_RESERVATIONS_DELETE_TIME- Automatically deletes an auto-created reservations at a specific time. Thespecified time must be an RFC3339 timestamp, which must be formatted as"YYYY-MM-DDTHH:MM:SSZ" where YYYY = year, MM = month, DD = day, HH = hours, MM =minutes, SS = seconds, and Z = time zone in Coordinated Universal Time (UTC).For example, specify 2021-11-20T07:00:00Z.
--auto-created-reservations-duration=AUTO_CREATED_RESERVATIONS_DURATION- Automatically deletes an auto-created reservations after a specified number ofdays, hours, minutes, or seconds. For example, specify 30m for 30 minutes, or1d2h3m4s for 1 day, 2 hours, 3 minutes, and 4 seconds. For more information, see$gcloud topic datetimes.
- Manage the commitment info properties
--commitment-name=COMMITMENT_NAME- Name of commitment covering the delivered reservation at the time of delivery offuture reservations. If not specified, it takes the name of the futurereservation.
--commitment-plan=COMMITMENT_PLAN- The plan for this commitment to be created, which determines duration anddiscount rate. The currently supported plans are TWELVE_MONTH (1 year), andTHIRTY_SIX_MONTH (3 years).
COMMITMENT_PLANmust be oneof:TWELVE_MONTH,THIRTY_SIX_MONTH. --previous-commitment-terms=PREVIOUS_COMMITMENT_TERMS- Applicable only if future reservation will deliver to an existing reservationwith a ramp plan. When set to EXTEND, all associated parent Committed UsedDiscount's end-date/term will be extended to the end-time of this futurereservation. Default is to extend previous commitment's time to the end_time ofthe reservation.
PREVIOUS_COMMITMENT_TERMSmust be (onlyone value is supported):EXTEND.
- Manage the properties of a shared reservation.
--share-setting=SHARE_SETTING- Specify if this future reservation is shared, and if so, the type of sharing. Ifyou omit this flag, this value is local by default.
SHARE_SETTINGmust be one of:local,projects. --share-with=PROJECT,[PROJECT,…]- If this future reservation is shared, provide a comma-separated list of projectsthat this future reservation is shared with. The list must contain project IDsor project numbers.
- GCLOUD WIDE FLAGS
- These flags are available to all commands:
--access-token-file,--account,--billing-project,--configuration,--flags-file,--flatten,--format,--help,--impersonate-service-account,--log-http,--project,--quiet,--trace-token,--user-output-enabled,--verbosity.Run
$gcloud helpfor details. - NOTES
- This command is currently in alpha and might change without notice. If thiscommand fails with API permission errors despite specifying the correct project,you might be trying to access an API with an invitation-only early accessallowlist. These variants are also available:
gcloudcomputefuture-reservationscreategcloudbetacomputefuture-reservationscreate
Except as otherwise noted, the content of this page is licensed under theCreative Commons Attribution 4.0 License, and code samples are licensed under theApache 2.0 License. For details, see theGoogle Developers Site Policies. Java is a registered trademark of Oracle and/or its affiliates.
Last updated 2025-10-28 UTC.