為地區標準叢集建立的預設節點集區包含九個節點 (每個區域三個),平均散佈到一個地區的三個區域。如果叢集使用公開節點,則會耗用九個 IP 位址。如有需要,您可以將節點數降為每個區域一個。新建立的 Cloud 帳單帳戶在每個區域只會授予八個 IP 位址,因此您可能需要根據區域叢集的大小,為區域使用中的 IP 位址申請調高配額。如果可用的使用中 IP 位址太少,叢集建立作業會失敗。
[[["容易理解","easyToUnderstand","thumb-up"],["確實解決了我的問題","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["難以理解","hardToUnderstand","thumb-down"],["資訊或程式碼範例有誤","incorrectInformationOrSampleCode","thumb-down"],["缺少我需要的資訊/範例","missingTheInformationSamplesINeed","thumb-down"],["翻譯問題","translationIssue","thumb-down"],["其他","otherDown","thumb-down"]],["上次更新時間:2025-07-29 (世界標準時間)。"],[],[],null,["# Regional clusters\n\n[Autopilot](/kubernetes-engine/docs/concepts/autopilot-overview) [Standard](/kubernetes-engine/docs/concepts/choose-cluster-mode)\n\n*** ** * ** ***\n\nThis page explains how regional clusters work in Google Kubernetes Engine (GKE).\n\nRegional clusters increase the availability of a cluster by replicating the\ncontrol plane across multiple zones in a\n[region](/compute/docs/regions-zones/regions-zones).\n\nThis configuration provides the following benefits:\n\n- **Resilience from single zone failure:** Regional clusters are available across a *region* rather than a single zone within a region. If a single zone becomes unavailable, your control plane is not impacted.\n- **Continuous control plane upgrades, control plane resizes, and reduced downtime from control plane failures**. With redundant replicas of the control plane, regional clusters provide higher availability of the Kubernetes API, so you can access your control plane even during upgrades.\n\nAlso, by default, regional clusters are created as [multi-zonal clusters](/kubernetes-engine/docs/concepts/configuration-overview#multi-zonal_clusters),\nso worker nodes are distributed across multiple zones in a region. This increases the\navailability of your workload, if you run sufficient replicas of the workload.\n\nGKE Autopilot clusters are always regional.\nIf you use GKE Standard, you can choose to create\nregional or zonal clusters. To learn about the different cluster\navailability types, see\n[Cluster availability](/kubernetes-engine/docs/concepts/configuration-overview#availability).\n\nIn regional clusters, including Autopilot clusters, the control plane\nis replicated across multiple zones of a region. GKE automatically\nreplicates nodes across zones in the same region. In Standard clusters and node\npools, you can optionally manually specify the zone(s) in which the nodes run.\nAll zones must be within the same region as the control plane.\n| **Note:** Use regional clusters to run your production workloads, as they generally offer higher availability than zonal clusters. For more information about region-specific considerations, see [Geography and regions](/docs/geography-and-regions#regions_and_zones).\n\nAfter creating a regional cluster, you cannot change it to a zonal cluster.\n\nHow regional clusters work\n--------------------------\n\nRegional clusters replicate the cluster's control plane and nodes across multiple zones\nwithin a single [region](/compute/docs/regions-zones/regions-zones#available).\nFor example, using the default configuration, a regional cluster in the\n`us-east1` region creates multiple replicas of the control plane in different `us-east1` zones and provisions nodes in three\n`us-east1` zones: `us-east1-b`, `us-east1-c`, and `us-east1-d`. In the event of\nan infrastructure outage, Autopilot workloads continue to run and\nGKE automatically rebalances nodes.\nIf you use Standard clusters, you must rebalance nodes manually or\nby using the\n[cluster autoscaler](/kubernetes-engine/docs/concepts/cluster-autoscaler).\n\n### Limitations\n\n- The default node pool created for regional Standard clusters consists\n of nine nodes (three per zone) spread evenly across three zones in a region.\n This consumes nine IP addresses for clusters using public nodes. You can\n reduce the number of nodes down to one per zone, if needed. Newly created\n Cloud Billing accounts are granted only eight IP addresses per region, so\n you may need to [request an increase in your quotas](/compute/quotas) for\n regional in-use IP addresses, depending on the size of your regional cluster.\n If you have too few available in-use IP addresses, cluster creation fails.\n\n- To run [GPUs](/kubernetes-engine/docs/concepts/gpus) in your regional cluster,\n choose a region that has at least one zone where the requested GPUs are\n available. You must use the\n [`--node-locations`](/kubernetes-engine/docs/how-to/gpus#create)\n flag when creating the node pool to specify the zone or zones containing the\n requested GPUs.\n\n If the region you choose doesn't have at least one zone where the requested GPUs are available,\n you might see an error like the following: \n\n\n ERROR: (gcloud.container.clusters.create) ResponseError: code=400, message=\n Accelerator type \"nvidia-l4\" does not exist in zone europe-west3-a.\n\n For a complete list of regions and zones where GPUs are available, see\n [GPUs on Compute Engine](/compute/docs/gpus).\n- Zones for Standard mode node pools must be in the same region as\n the cluster's control plane. If you need to, you can\n [change a cluster's zones](/kubernetes-engine/docs/how-to/managing-clusters#add_or_remove_zones),\n which causes all new and existing nodes to span those zones.\n\n### Pricing\n\nAll Autopilot clusters are regional, and are subject to the\n[Autopilot pricing model](/kubernetes-engine/pricing#autopilot_mode).\n\nIn Standard mode, regional clusters require more of your project's\n[regional quotas](/kubernetes-engine/quotas) than a similar zonal or multi-zonal\ncluster. Ensure that you understand your quotas and\n[Standard pricing](/kubernetes-engine/pricing#standard_mode)\nbefore using regional clusters. If you encounter an\n`Insufficient regional quota to satisfy request for resource` error, your\nrequest exceeds your available quota in the current region.\n\nAlso, you are charged for node-to-node traffic across zones. For\nexample, if a workload running in one zone needs to communicate with a\nworkload in a different zone, the cross-zone traffic incurs cost. For more\ninformation, see\n[Egress between zones in the same region (per GB)](/vpc/network-pricing#general)\nin the Compute Engine pricing page.\n\nPersistent storage in regional clusters\n---------------------------------------\n\nZonal persistent disks are zonal resources and regional persistent disks are\nmulti-zonal resources. When adding\n[persistent storage](/kubernetes-engine/docs/how-to/stateful-apps#requesting_persistent_storage_in_a_statefulset)\nunless a zone is specified, GKE assigns the disk\nto a single, random zone. To learn how to control the zones, see\n[Zones in persistent disks](/kubernetes-engine/docs/concepts/persistent-volumes#pd-zones).\n\nAutoscaling regional clusters\n-----------------------------\n\nKeep the following considerations in mind when using the\n[cluster autoscaler](/kubernetes-engine/docs/concepts/cluster-autoscaler)\nto automatically scale node pools in regional Standard mode\nclusters.\n\nYou can also learn more about\n[Autoscaling limits](/kubernetes-engine/docs/concepts/cluster-autoscaler#autoscaling_limits)\nfor regional clusters or about how Cluster Autoscaler\n[balances across zones](/kubernetes-engine/docs/concepts/cluster-autoscaler#balancing_across_zones).\n\nThese considerations only apply to Standard mode clusters\nwith the cluster autoscaler.\n\n### Overprovisioning scaling limits\n\nTo maintain capacity in the unlikely event of zonal failure, you can allow\nGKE to overprovision your scaling limits, to ensure a\nminimum level of availability even when some zones are unavailable.\n\nFor example, if you overprovision a three-zone cluster to 150% (50%\nexcess capacity), you can ensure that 100% of traffic is routed to available\nzones if one-third of the cluster's capacity is lost. In the preceding example, you\nwould accomplish this by specifying a maximum of six nodes per zone rather than\nfour. If one zone fails, the cluster scales to 12 nodes in the remaining\nzones.\n\nSimilarly, if you overprovision a two-zone cluster to 200%, you can ensure\nthat 100% of traffic is rerouted if half of the cluster's capacity is lost.\n\nYou can learn more about the\n[cluster autoscaler](/kubernetes-engine/docs/concepts/cluster-autoscaler)\nor read the [FAQ for autoscaling](https://github.com/kubernetes/autoscaler/blob/master/cluster-autoscaler/FAQ.md) in the Kubernetes documentation.\n\nWhat's next\n-----------\n\n- [Create a regional cluster](/kubernetes-engine/docs/how-to/creating-a-regional-cluster).\n- Learn more about the different [types of clusters](/kubernetes-engine/docs/concepts/types-of-clusters).\n- [Learn more about node pools](/kubernetes-engine/docs/concepts/node-pools).\n- [Learn more about cluster architecture](/kubernetes-engine/docs/concepts/cluster-architecture)."]]