This page describes how Google Kubernetes Engine (GKE) Autopilot manages the values of workload resource requests, such as CPU, memory, or ephemeral storage. This page includes the following information, which you can use to plan efficient, stable, and cost-effective workloads:
This page is for Operators and Developers who provision and configure cloud resources, and deploy workloads. To learn more about common roles and example tasks that we reference in Google Cloud content, see Common GKE user roles and tasks.
Before reading this page, ensure that you're familiar with Kubernetes resource management concepts.
Overview of resource requests in AutopilotAutopilot uses the resource requests that you specify in your workload configuration to configure the nodes that run your workloads. Autopilot enforces minimum and maximum resource requests based on the compute class or the hardware configuration that your workloads use. If you don't specify requests for some containers, Autopilot assigns default values to let those containers run correctly.
When you deploy a workload in an Autopilot cluster, GKE validates the workload configuration against the allowed minimum and maximum values for the selected compute class or hardware configuration (such as GPUs). If your requests are less than the minimum, Autopilot automatically modifies your workload configuration to bring your requests within the allowed range. If your requests are greater than the maximum, Autopilot rejects your workload and displays an error message.
The following list summarizes the categories of resource requests:
In Autopilot, you request resources in your Pod specification. The supported minimum and maximum resources that you can request change based on the hardware configuration of the node on which the Pods run. To learn how to request specific hardware configurations, refer to the following pages:
Default resource requestsIf you don't specify resource requests for some containers in a Pod, Autopilot applies default values. These defaults are suitable for many smaller workloads.
Note: We recommend that you explicitly set your resource requests for each container to meet your application requirements, as these default values might not be sufficient, or optimal.Additionally, Autopilot applies the following default resource requests regardless of the selected compute class or hardware configuration:
Containers in DaemonSets
All other containers
For more information about Autopilot cluster limits, see Quotas and limits.
Default requests for compute classesAutopilot applies the following default values to resources that are not defined in the Pod specification for Pods that run on compute classes. If you only set one of the requests and leave the other blank, GKE uses the CPU:memory ratio defined in the Minimum and maximum requests section to set the missing request to a value that complies with the ratio.
Compute class Resource Default request General-purpose (default) CPU 0.5 vCPU Memory 2 GiB Accelerator See the Default resources for accelerators section. Balanced CPU 0.5 vCPU Memory 2 GiB Performance CPUIn version 1.29.4-gke.1427000 and later, Autopilot doesn't enforce default requests for accelerators. To learn more, see Pricing.
The following table describes the default values that GKE assigns to Pods that don't specify values in the requests field of the Pod specification. This table applies to Pods that run on versions earlier than 1.29.4-gke.1427000 that use the Accelerator compute class, which is the recommended way to run accelerators in Autopilot clusters.
nvidia-b200 No default requests enforced. NVIDIA H200 (141GB) GPUs
nvidia-h200-141gb No default requests enforced. NVIDIA H100 Mega (80GB) GPUs
nvidia-h100-mega-80gb CPU
nvidia-h100-80gb CPU
nvidia-tesla-a100 CPU
nvidia-a100-80gb CPU
nvidia-l4 CPU
nvidia-tesla-t4 CPU
tpu-v6e-slice (single-host) CPU All topologies: 1 mCPU Memory All topologies: 1 MiB TPU Trillium (v6e)
tpu-v6e-slice (multi-host) CPU All topologies: 1 mCPU Memory All topologies: 1 MiB TPU v5e
tpu-v5-lite-podslice (multi-host) CPU All topologies: 1 mCPU Memory All topologies: 1 MiB TPU v5p
tpu-v5p-slice CPU All topologies: 1 mCPU Memory All topologies: 1 MiB TPU v4
tpu-v4-podslice CPU All topologies: 1 mCPU Memory All topologies: 1 MiB Supported GPUs without the Accelerator compute class
If you don't use the Accelerator compute class, only the following GPUs are supported. The default resource requests for these GPUs are the same as in the Accelerator compute class:
The total resources requested by your deployment configuration should be within the supported minimum and maximum values that Autopilot allows. The following conditions apply:
Ephemeral storage requests:
Ephemeral storage uses the VM boot disk unless your nodes have local SSDs attached.
Compute hardware that includes local SSDs like A100 (80GB) GPUs, H100 (80GB) GPUs, or the Z3 machine series support a maximum request that's equal to the size of the local SSD minus any system overhead. For information about this system overhead, see Ephemeral storage backed by local SSDs.
In GKE version 1.29.3-gke.1038000 and later, Performance class Pods and hardware accelerator Pods support a maximum ephemeral storage request of 56 Ti unless the hardware includes local SSDs.
In all other Autopilot Pods regardless of the GKE version, the total ephemeral storage request across all of the containers in the Pod must be between 10 MiB and 10 GiB unless otherwise specified.
For larger volumes, use generic ephemeral volumes, which provide equivalent functionality and performance to ephemeral storage but with significantly more flexibility as they can be used with any GKE storage option. For example, the maximum size for a generic ephemeral volume using pd-balanced is 64 TiB.
For DaemonSet Pods, the minimum resource requests are as follows:
To check whether your cluster supports bursting, see Bursting availability in GKE.
If your cluster supports bursting, Autopilot doesn't enforce 0.25 vCPU increments for your Pod CPU requests. If your cluster doesn't support bursting, Autopilot rounds up your CPU requests to the nearest 0.25 vCPU. To check whether your cluster supports bursting, see Bursting availability in GKE.
The CPU:memory ratio must be within the allowed range for the selected compute class or hardware configuration. If your CPU:memory ratio is outside the allowed range, Autopilot automatically increases the smaller resource. For example, if you request 1 vCPU and 16 GiB of memory (1:16 ratio) for Pods running on the Scale-Out class, Autopilot increases the CPU request to 4 vCPUs, which changes the ratio to 1:4.
The following table describes the minimum, maximum, and allowed CPU-to-memory ratio for each compute class that Autopilot supports:
Compute class CPU:memory ratio (vCPU:GiB) Resource Minimum Maximum General-purpose (default) Between 1:1 and 1:6.5 CPUThe value depends on whether your cluster supports bursting, as follows:
To check whether your cluster supports bursting, see Bursting availability in GKE.
30 vCPU MemoryThe value depends on whether your cluster supports bursting, as follows:
To check whether your cluster supports bursting, see Bursting availability in GKE.
110 GiB Accelerator See Minimums and maximums for accelerators Balanced Between 1:1 and 1:8 CPU 0.25 vCPU222 vCPU
If minimum CPU platform selected:
851 GiB
If minimum CPU platform selected:
In GKE version 1.29.3-gke.1038000 and later, you can specify a maximum ephemeral storage request of 56 Ti.
The C4D machine series is available with version 1.33.0-gke.1439000 or later and supports requests of up to 56 Ti with or without Local SSD.
For versions earlier than 1.29.3-gke.1038000, the following limits apply:
arm64: 43 vCPUamd64: 54 vCPUarm64: 172 GiBamd64: 216 GiBTo learn how to request compute classes in your Autopilot Pods, refer to Choose compute classes for Autopilot Pods.
Minimums and maximums for acceleratorsThe following sections describe the minimum, maximum, and allowed CPU-to-memory ratio for Pods that use hardware accelerators like GPUs and TPUs.
Unless specified, the maximum ephemeral storage supported is 122 GiB in versions 1.28.6-gke.1369000 or later, and 1.29.1-gke.1575000 or later. For earlier versions, the maximum ephemeral storage supported is 10 GiB.
Minimums and maximums for the Accelerator compute classThe following table shows the minimum and maximum resource requests for Pods that use the Accelerator compute class, which is the recommended way to run accelerators with GKE Autopilot clusters. In the Accelerator compute class, GKE doesn't enforce CPU-to-memory request ratios.
Accelerator type Resource Minimum Maximum NVIDIA B200nvidia-B200 CPU No minimum requests enforced
nvidia-h200-141gb CPU No minimum requests enforced
nvidia-h100-mega-80gb CPU
nvidia-h100-80gb CPU
nvidia-tesla-a100 CPU 0.001 vCPU
The sum of CPU requests of all DaemonSets that run on an A100 GPU node must not exceed 2 vCPU.
Memory 1 MiBThe sum of memory requests of all DaemonSets that run on an A100 GPU node must not exceed 14 GiB.
NVIDIA A100 (80GB)nvidia-a100-80gb CPU 0.001 vCPU
The sum of CPU requests of all DaemonSets that run on an A100 (80GB) GPU node must not exceed 2 vCPU.
Memory 1 MiBThe sum of memory requests of all DaemonSets that run on an A100 (80GB) GPU node must not exceed 14 GiB.
Ephemeral storage 512 MiBnvidia-l4 CPU 0.001 vCPU
The sum of CPU requests of all DaemonSets that run on an L4 GPU node must not exceed 2 vCPU.
Memory 1 MiBThe sum of memory requests of all DaemonSets that run on an L4 GPU node must not exceed 14 GiB.
NVIDIA Tesla T4nvidia-tesla-t4 CPU 0.001 vCPU
tpu-v5-lite-podslice CPU 0.001 vCPU
tpu-v5p-slice CPU 0.001 vCPU 280 vCPU Memory 1 MiB 448 GiB Ephemeral storage 10 MiB 56 TiB TPU v4
tpu-v4-podslice CPU 0.001 vCPU 240 vCPU Memory 1 MiB 407 GiB Ephemeral storage 10 MiB 56 TiB Note: All A100 (80GB) GPU nodes use local SSDs for node boot disks at fixed sizes based on the number of GPUs. You're billed separately for the attached Local SSDs. For details, see Autopilot pricing. A100 (40GB) GPUs don't use local SSDs for node boot disks.
To learn how to request GPUs in your Autopilot Pods, refer to Deploy GPU workloads in Autopilot.
Minimums and maximums for GPUs without a compute classThe following table shows the minimum and maximum resource requests for Pods that don't use the Accelerator compute class:
GPU type CPU:memory ratio (vCPU:GiB) Resource Minimum Maximum NVIDIA A100 (40GB)nvidia-tesla-a100 Not enforced CPU
The sum of CPU requests of all DaemonSets that run on an A100 GPU node must not exceed 2 vCPU.
MemoryThe sum of memory requests of all DaemonSets that run on an A100 GPU node must not exceed 14 GiB.
NVIDIA A100 (80GB)nvidia-a100-80gb Not enforced CPU
The sum of CPU requests of all DaemonSets that run on an A100 (80GB) GPU node must not exceed 2 vCPU.
MemoryThe sum of memory requests of all DaemonSets that run on an A100 (80GB) GPU node must not exceed 14 GiB.
Ephemeral storagenvidia-l4
The sum of CPU requests of all DaemonSets that run on an L4 GPU node must not exceed 2 vCPU.
MemoryThe sum of memory requests of all DaemonSets that run on an L4 GPU node must not exceed 14 GiB.
NVIDIA Tesla T4nvidia-tesla-t4 Between 1:1 and 1:6.25 CPU 0.5 vCPU
To learn how to request GPUs in your Autopilot Pods, refer to Deploy GPU workloads in Autopilot.
Resource requests for workload separation and extended durationAutopilot lets you manipulate Kubernetes scheduling and eviction behavior using methods such as the following:
If your specified requests are less than the minimums, the behavior of Autopilot changes based on the method that you used, as follows:
The following table describes the default requests and the minimum resource requests that you can specify. If a configuration or compute class isn't in this table, Autopilot doesn't enforce special minimum or default values.
Compute class Resource Default Minimum General-purpose CPU 0.5 vCPU 0.5 vCPU Memory 2 GiB 0.5 GiB Balanced CPU 2 vCPU 1 vCPU Memory 8 GiB 4 GiB Scale-Out CPU 0.5 vCPU 0.5 vCPU Memory 2 GiB 2 GiB Init containersInit containers run in serial and must complete before the application containers start. If you don't specify resource requests for your Autopilot init containers, GKE allocates the total resources available to the Pod to each init container. This behavior is different than in GKE Standard, where each init container can use any unallocated resources available on the node on which the Pod is scheduled.
Unlike application containers, GKE recommends that you don't specify resource requests for Autopilot init containers, so that each container gets the full resources available to the Pod. If you request less resources than the defaults, you constrain your init container. If you request more resources than the Autopilot defaults, you might increase your bill for the lifetime of the Pod.
Setting resource limits in AutopilotKubernetes lets you set both requests and limits for resources in your Pod specification. The behavior of your Pods changes depending on whether your limits are different than your requests, as described in the following table:
requests equal to limits Pods use the Guaranteed QoS class. Note: Ephemeral storage limits must always be explicitly set equal to requests. GKE modifies your Pods to enforce this rule. requests set, limits not set
The behavior depends on whether your cluster supports bursting, as follows:
limits equal to the requestsTo check whether your cluster supports bursting, see Bursting availability in GKE.
requests not set, limits set Autopilot sets requests to the value of limits, which is the default Kubernetes behavior.
Before:
resources:
limits:
cpu: "400m"
After:
resources:
requests:
cpu: "400m"
limits:
cpu: "400m"requests less than limits
The behavior depends on whether your cluster supports bursting, as follows:
limits.limits equal to the requestsTo check whether your cluster supports bursting, see Bursting availability in GKE.
requests greater than limits Autopilot sets requests to the value of limits.
Before:
resources:
requests:
cpu: "450m"
limits:
cpu: "400m"
After:
resources:
requests:
cpu: "400m"
limits:
cpu: "400m"requests not set, limits not set
Autopilot sets requests to the default values for the compute class or hardware configuration.
The behavior for limits depends on whether your cluster supports bursting, as follows:
limits.limits equal to the requestsTo check whether your cluster supports bursting, see Bursting availability in GKE.
In most situations, set adequate resource requests and equal limits for your workloads.
For workloads that temporarily need more resources than their steady-state, like during boot up or during higher traffic periods, set your limits higher than your requests to let the Pods burst. For details, see Configure Pod bursting in GKE.
Automatic resource management in AutopilotIf your specified resource requests for your workloads are outside of the allowed ranges, or if you don't request resources for some containers, Autopilot modifies your workload configuration to comply with the allowed limits. Autopilot calculates resource ratios and the resource scale up requirements after applying default values to containers with no request specified.
By default, when Autopilot automatically scales a resource up to meet a minimum or default resource value, GKE allocates the extra capacity to the first container in the Pod manifest. In GKE version 1.27.2-gke.2200 and later, you can tell GKE to allocate the extra resources to a specific container by adding the following to the annotations field in your Pod manifest:
autopilot.gke.io/primary-container: "CONTAINER_NAME"
Replace CONTAINER_NAME with the name of the container.
The following example scenario shows how Autopilot modifies your workload configuration to meet the requirements of your running Pods and containers.
Single container with < 0.05 vCPU Container number Original request Modified request 1 CPU: 30 mCPUIn this example, the memory is too low for the amount of CPU (1 vCPU:1 GiB minimum). The minimum allowed ratio for CPU to memory is 1:1. If the ratio is lower than that, the memory request is increased.
Container number Original request Modified request 1 CPU: 4 vCPURetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4