[[["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：2025-09-04 (世界標準時間)。"],[],[],null,["# Identify underprovisioned and overprovisioned GKE clusters\n\n[Standard](/kubernetes-engine/docs/concepts/choose-cluster-mode)\n\n*** ** * ** ***\n\nThis page explains how to identify underprovisioned and overprovisioned\nGoogle Kubernetes Engine (GKE) clusters. GKE provides insights and\nrecommendations for cost optimization scenarios such as overprovisioned clusters\nand idle clusters, and reliability improvement scenarios such as\nunderprovisioned clusters. GKE provides corresponding\nrecommendations to scale up, scale down, or delete the clusters. For idle\nclusters, see [Identify idle GKE\nclusters](/kubernetes-engine/docs/how-to/idle-clusters).\n\nAfter you verify that the identified clusters would benefit from the\nrecommendation to scale up or down, you can make the recommended change to save\ncosts or increase the reliability of your cluster. If possible, the\nrecommendation includes projected monthly savings or cost. For more information,\nsee [Understand cost or savings estimates](#cost-or-savings-estimate).\n\nGKE doesn't provide these insights for Autopilot\nclusters, which incur minimal operational costs because you only pay for the\nresources that your workloads request. For more information, see\n[Autopilot\nPricing](/kubernetes-engine/docs/concepts/autopilot-overview#pricing).\n\nGKE monitors your clusters and delivers guidance to optimize your\nusage through [Active Assist](/recommender/docs/whatis-activeassist), a\nservice that provides recommenders that generate insights and recommendations\nfor using resources on Google Cloud. For more information about how to manage\ninsights and recommendations, see [Optimize your usage of GKE\nwith insights and\nrecommendations](/kubernetes-engine/docs/how-to/optimize-with-recommenders).\n\nGet insights and recommendations for underprovisioned and overprovisioned clusters\n----------------------------------------------------------------------------------\n\nGKE surfaces these insights and recommendations in the following\nlocations in the Google Cloud console:\n\n- **Kubernetes Clusters** page, in the following locations:\n - In the **Kubernetes clusters** list, in the **Notifications** column for the applicable clusters\n - Notification banners on the **Clusters** page for a specific cluster\n- [FinOps hub](/billing/docs/how-to/finops-hub)\n\nThe recommendations have the following titles in the **Kubernetes Clusters**\npage:\n\n- **Overprovisioned clusters**: \"Decrease cluster resources to reduce costs\"\n- **Underprovisioned clusters**: \"Increase cluster resources to improve reliability\"\n\nYou can also receive these insights and recommendations through the\nGoogle Cloud CLI or the Recommender API, using the `CLUSTER_UNDERPROVISIONED`\nand `CLUSTER_OVERPROVISIONED` subtypes.\n\nFollow the instructions to [view insights and\nrecommendations](/kubernetes-engine/docs/how-to/optimize-with-recommenders#view-insights-recs).\n\nAfter you identify underprovisioned or overprovisioned clusters, see the\n[considerations when rightsizing clusters](#rightsize-cluster).\n\nHow GKE identifies underprovisioned and overprovisioned clusters\n----------------------------------------------------------------\n\nThe following table describes the signals that GKE uses for\nidentifying underprovisioned and overprovisioned clusters that can be scaled up\nor down, and the threshold for each signal. Additionally, this table shows the\naction we recommend that you take in this scenario.\n\nGKE doesn't send recommendations for clusters that were created\nless than 30 days ago.\n\nUnderstand cost or savings estimates\n------------------------------------\n\nIf possible, GKE's recommendation includes an estimate that\nprojects the monthly cost or savings if you rightsized the cluster. This\nestimate is derived from the cluster costs over the past 30 days.\n\nAny estimated costs or savings are projections based on previous spending, and\nare not a guarantee of future cost or savings.\n\nTo see these estimates, ensure that you have the required\n`billing.accounts.getSpendingInformation` permission to get spending\ninformation. For details, see [Cloud Billing\naccess](/billing/docs/how-to/billing-access#billing-access).\n\nTo get more information about the cost of all of your GKE\nclusters, including a more granular breakdown based on namespaces and workloads,\nsee [Get key spending insights for your GKE resource allocation\nand cluster costs](/kubernetes-engine/docs/how-to/cost-allocations).\n\nFor more information about the costs of running a GKE cluster,\nsee [GKE pricing](/kubernetes-engine/pricing).\n\nConsiderations when rightsizing clusters\n----------------------------------------\n\nBefore you follow a recommendation to scale up or down a cluster, consider the\nfollowing:\n\n- Review the resource utilization of applications running on your cluster to see how they're performing, and if they're using more or less CPU and memory than expected. For instructions, see [Analyze resource\n requests](/kubernetes-engine/docs/how-to/vertical-pod-autoscaling#get-resource-recommendations).\n- Batch processing workloads might intentionally maintain high utilization of cluster resources for cost efficiency. If the allocated cluster resources are sufficient for the batch jobs running on the cluster, you don't need to scale up the highly utilized cluster, which was identified as underprovisioned.\n\nImplement the recommendation to rightsize a cluster\n---------------------------------------------------\n\nReview the following to understand how you can adjust the size of a cluster to\nbetter match your resource utilization.\n\n### Rightsize an underprovisioned cluster\n\nTo implement the recommendation to minimize the risk of reliability by\nrightsizing an underprovisioned cluster, increase resources on the cluster. You\ncan do so by taking some of the following actions:\n\n- Enable [cluster\n autoscaler](/kubernetes-engine/docs/concepts/cluster-autoscaler) and [node\n auto-provisioning](/kubernetes-engine/docs/concepts/node-auto-provisioning), or adjust the settings to allow for greater scaling up.\n- Horizontally scale up your cluster by increasing the number of nodes. Follow the instructions to [horizontally scale by changing the node\n count](/kubernetes-engine/docs/how-to/node-pools#resizing_a_node_pool).\n- Choose a larger machine type for your node pools. Follow the instructions to [vertically scale by changing the node machine\n attributes](/kubernetes-engine/docs/how-to/node-pools#change-machine-attributes).\n- Monitor and review the CPU and memory resource usage of applications that run on your cluster. See if you can scale down applications. For instructions about monitoring resource usage, see [Analyze resource\n requests](/kubernetes-engine/docs/how-to/vertical-pod-autoscaling#get-resource-recommendations).\n\nWhen you implement this recommendation, you ensure that your cluster remains\nreliable because it has the appropriate amount of resources for its\napplications.\n\n### Rightsize an overprovisioned cluster\n\nTo implement the recommendation to save costs by rightsizing an overprovisioned\ncluster, decrease resources on the cluster. Adjust cluster CPU and memory\nallocations to match your workload needs. You can do so by taking some of the\nfollowing actions:\n\n- Adjust [cluster\n autoscaler](/kubernetes-engine/docs/concepts/cluster-autoscaler) and [node\n auto-provisioning](/kubernetes-engine/docs/concepts/node-auto-provisioning) to more aggressively scale down underutilized resources.\n- Horizontally scale down your cluster by decreasing the number of nodes. Follow the instructions to [horizontally scale by changing the node\n count](/kubernetes-engine/docs/how-to/node-pools#resizing_a_node_pool).\n- Choose a smaller machine type for your node pools. Follow the instructions to [vertically scale by changing the node machine\n attributes](/kubernetes-engine/docs/how-to/node-pools#change-machine-attributes).\n- Monitor and review the CPU and memory resource usage of applications that run on your cluster. See if you can scale up applications. For instructions about monitoring resource usage, see [Analyze resource\n requests](/kubernetes-engine/docs/how-to/vertical-pod-autoscaling#get-resource-recommendations).\n\nWhen you implement this recommendation, you ensure that you're not using more\nresources than necessary to run your cluster's applications.\n\nWhat's next\n-----------\n\n- [View cost-related optimization\n metrics](/kubernetes-engine/docs/how-to/cost-optimization-metrics)\n- [Reducing costs by scaling down GKE clusters during off-peak\n hours](/kubernetes-engine/docs/tutorials/reducing-costs-by-scaling-down-gke-off-hours)\n- [Optimize your usage of GKE with insights and\n recommendations](/kubernetes-engine/docs/how-to/optimize-with-recommenders)\n- [Best practices for running cost-optimized Kubernetes applications on\n GKE](/architecture/best-practices-for-running-cost-effective-kubernetes-applications-on-gke)\n- [5 GKE features to help you optimize your\n clusters](/blog/products/containers-kubernetes/boost-your-gke-game-with-these-tips-and-tutorials)"]]
