- NAME
-
- gcloud alpha compute advice calendar-mode - recommends the optimal time window and zone for Future Reservations
- SYNOPSIS
-
-
gcloud alpha compute advice calendar-mode
([--chip-count
=CHIP_COUNT
--tpu-version
=TPU_VERSION
:--workload-type
=WORKLOAD_TYPE
] |--machine-type
=MACHINE_TYPE
--vm-count
=VM_COUNT
) [--duration-range
=[max
=MAX
],[min
=MIN
]] [--end-time-range
=[from
=FROM
],[to
=TO
]] [--location-policy
=[ZONE
=POLICY
,…]] [--region
=REGION
] [--start-time-range
=[from
=FROM
],[to
=TO
]] [GCLOUD_WIDE_FLAG …
]
-
- DESCRIPTION
-
(ALPHA)
Recommends the optimal time window and zone for Future Reservations. - EXAMPLES
-
To request an advice for a future reservation of 8 a3-highgpu-8g VMs, run:
gcloud alpha compute advice calendar-mode --region=fake-region --machine-type=a3-megagpu-8g --vm-count=8 --duration-range=min=7d,max=14d --start-time-range=from=2025-02-20,to=2025-03-25 --end-time-range=from=2025-02-20,to=2025-03-25
To request advice for a future reservation of 512 v5e TPUs, run:
gcloud alpha compute advice calendar-mode --region=fake-region --location-policy=fake-zone-1=DENY,fake-zone-2=ALLOW --tpu-version=V5E --chip-count=16 --workload-type=BATCH --duration-range=min=30d,max=90d --start-time-range=from=2025-02-25,to=2025-06-25 --end-time-range=from=2025-02-25,to=2025-06-25
- REQUIRED FLAGS
-
Exactly one of these must be specified:Specify the properties of the resources that you want to view the availability of.
You must define the version and number of TPUs to reserve.
--chip-count
=CHIP_COUNT
-
The number of chips to check for availability.
This flag argument must be specified if any of the other arguments in this group are specified.
--tpu-version
=TPU_VERSION
-
The version of Cloud TPU to reserve.
TPU_VERSION
must be one of:V5E
-
Cloud TPU v5e Lite
V5P
- Cloud TPU v5p
V6E
- Cloud TPU v6e
--workload-type
=WORKLOAD_TYPE
-
Type of the workload that will be running on the reserved TPUs.
WORKLOAD_TYPE
must be one of:BATCH
- Reserved resources will be optimized for BATCH workloads that handle large amounts of data in single or multiple operations, such as ML training workloads.
SERVING
- Reserved resources will be optimized for SERVING workloads that handle concurrent requests and require minimal network latency, such as ML inference.
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 fixed
amount of memory. You can also specify a custom machine type by using the
pattern
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 group are specified.
--vm-count
=VM_COUNT
-
The number of instances to check for availability.
This flag argument must be specified if any of the other arguments in this group are specified.
- OPTIONAL FLAGS
-
--duration-range
=[max
=MAX
],[min
=MIN
]-
A duration range for the duration of the Future Reservation. Defined as a list
of key-value pairs.
The key is either "min" or "max", and the value is a duration in seconds. For example, specify
30m
for a duration of 30 minutes or specify1d2h3m4s
for a duration of 1 day, 2 hours, 3 minutes, and 4 seconds.See <a href="https://cloud.google.com/sdk/gcloud/reference/topic/datetimes"> $ gcloud topic datetimes</a> for information on duration format.
min
- The minimum duration of the Future Reservation.
max
-
The maximum duration of the Future Reservation.
Example:
gcloud compute advice calendar-mode --duration-range=min=24h,max=72h
--end-time-range
=[from
=FROM
],[to
=TO
]-
A time range for the end time of the Future Reservation. Defined as a list of
key-value pairs.
The key is either "from" or "to", and the value is a datetime. See <a href="https://cloud.google.com/sdk/gcloud/reference/topic/datetimes"> $ gcloud topic datetimes</a> for information on time formats.
from
- The earliest possible end time for the reservation.
to
-
The latest possible end time for the reservation.
Example:
gcloud compute advice calendar-mode --end-time-range=from=2024-08-01T00:00:00Z,to=2024-08-02T00:00:00Z
--location-policy
=[ZONE
=POLICY
,…]-
Policy for which zones to include or exclude when looking for the optimal time
window and zone for Future Reservations within a region. Policy is defined as a
list of key-value pairs, with the key being the zone name, and value being the
applied policy. Available policies are
allow
anddeny
. Default for zones if left unspecified isallow
.Example:
gcloud compute advice calendar-mode --location-policy=us-central1-a=allow,us-central1-b=deny
--region
=REGION
-
Region of the to None. If not specified, you might be prompted to select a
region (interactive mode only).
To avoid prompting when this flag is omitted, you can set the
property:compute/region
gcloud config set compute/region REGION
A list of regions can be fetched by running:
gcloud compute regions list
To unset the property, run:
gcloud config unset compute/region
Alternatively, the region can be stored in the environment variable
.CLOUDSDK_COMPUTE_REGION
--start-time-range
=[from
=FROM
],[to
=TO
]-
A time range for the start time of the Future Reservation. Defined as a list of
key-value pairs.
The key is either "from" or "to", and the value is a datetime. See <a href="https://cloud.google.com/sdk/gcloud/reference/topic/datetimes"> $ gcloud topic datetimes</a> for information on time formats.
from
- The earliest possible start time for the reservation.
to
-
The latest possible start time for the reservation.
Example:
gcloud compute advice calendar-mode --start-time-range=from=2024-08-01T00:00:00Z,to=2024-08-02T00:00:00Z
- 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 help
for details. - NOTES
-
This command is currently in alpha and might change without notice. If this
command fails with API permission errors despite specifying the correct project,
you might be trying to access an API with an invitation-only early access
allowlist. This variant is also available:
gcloud beta compute advice calendar-mode
gcloud alpha compute advice calendar-mode
Except as otherwise noted, the content of this page is licensed under the Creative Commons Attribution 4.0 License, and code samples are licensed under the Apache 2.0 License. For details, see the Google Developers Site Policies. Java is a registered trademark of Oracle and/or its affiliates.
Last updated 2025-05-06 UTC.