[[["易于理解","easyToUnderstand","thumb-up"],["解决了我的问题","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["很难理解","hardToUnderstand","thumb-down"],["信息或示例代码不正确","incorrectInformationOrSampleCode","thumb-down"],["没有我需要的信息/示例","missingTheInformationSamplesINeed","thumb-down"],["翻译问题","translationIssue","thumb-down"],["其他","otherDown","thumb-down"]],["最后更新时间 (UTC):2025-08-28。"],[[["\u003cp\u003eCompute Engine automatically generates reservation recommendations, based on the previous seven days of usage, to identify idle or underutilized on-demand reservations.\u003c/p\u003e\n"],["\u003cp\u003eThese recommendations help users optimize resource usage, potentially leading to cost savings by allowing the modification or deletion of unnecessary reservations.\u003c/p\u003e\n"],["\u003cp\u003eUsers can customize the recommendation algorithm by adjusting the observation period (from 7 to 30 days) and setting utilization thresholds, ensuring that recommendations align with their specific workloads and needs.\u003c/p\u003e\n"],["\u003cp\u003eThere are no costs associated with using these reservation recommendations, and they can be viewed and applied directly through the Compute Engine interface.\u003c/p\u003e\n"],["\u003cp\u003eThe generated recommendations are not applicable to reservations attached to committed use discounts (CUDs) or virtual machine (VM) instances with TPUs.\u003c/p\u003e\n"]]],[],null,["# Reservation recommendations\n\n*** ** * ** ***\n\n|\n| **Preview**\n|\n|\n| This product or feature is subject to the \"Pre-GA Offerings Terms\" in the General Service Terms section\n| of the [Service Specific Terms](/terms/service-terms#1).\n|\n| Pre-GA products and features are available \"as is\" and might have limited support.\n|\n| For more information, see the\n| [launch stage descriptions](/products#product-launch-stages).\n\nThis page explains how Compute Engine generates reservation recommendations\nand the parameters to configure them.\n\nCompute Engine provides reservation recommendations to help you identify\nidle or underutilized on-demand reservations for the previous seven days so that\nyou can [modify](/compute/docs/instances/reservations-modify) or\n[delete the reservations](/compute/docs/instances/reservations-delete).\n\nCompute Engine generates recommendations automatically based on system\nmetrics gathered by the Cloud Monitoring service. You can configure\nreservations recommendations to receive more or fewer recommendations.\n\n- To identify these recommendations and take action, see [View and apply idle reservation recommendations](/compute/docs/instances/view-and-apply-idle-reservation-recommendations) or [View and apply underutilized reservation recommendations](/compute/docs/instances/view-and-apply-underutilized-reservation-recommendations).\n- To configure these recommendations, see [Configure idle reservation recommendations](/compute/docs/instances/configure-idle-reservation-recommendations) or [Configure underutilized reservation recommendations](/compute/docs/instances/configure-underutilized-reservation-recommendations).\n- For an overview of Compute Engine reservations, see [About reservations](/compute/docs/instances/reservations-overview).\n\nPricing\n-------\n\nThere are no costs associated with using idle reservation recommendations. Using\nrecommendations to reduce your resource usage can result in cost savings. The\ndisplayed cost savings estimate is your potential monthly savings if you adjust\nyour VM reservation to match your actual usage. For example, if you\nreserved 8 VMs but consistently use only 1, you see the cost savings of\ndownsizing your reservation to 1 VM.\n\nLimitations\n-----------\n\nIdle and underutilized reservation recommendations are not available for the\nfollowing reservations:\n\n- On-demand reservations that are attached to committed use discounts (CUDs)\n- On-demand reservations for virtual machine (VM) instances with TPUs\n\nHow detection of idle and underutilized reservations works\n----------------------------------------------------------\n\nReservation recommendations for Compute Engine are based on\nhistorical usage metrics. By default, the historical observation period is\nthe previous 7 days. By changing the default observation period, you can\ncustomize the recommendations that you receive.\n\nTo generate recommendations, the algorithm considers reservations that accrue\ncosts, but aren't associated with an active Compute Engine resource\nfor the previous 7 days.\n\n### Frequency of recommendations\n\nAfter a reservation is created and you haven't consumed any resources for at\nleast 7 days, Compute Engine begins generating recommendations for it.\nNew recommendations are generated once per day.\n\nCustomize recommendations\n-------------------------\n\nCompute Engine lets you customize the recommendations you receive for\nyour project by changing the configuration used by the recommendation algorithm.\nIn particular, by changing the default observation period, you can receive\nrecommendations that better fit your workloads, applications,\nand infrastructure needs.\n\nTo learn how to modify the configuration for your project,\nsee the following:\n\n- [Configure idle reservation recommendations](/compute/docs/instances/configure-idle-reservation-recommendations)\n- [Configure underutilized reservation recommendations](/compute/docs/instances/configure-underutilized-reservation-recommendations).\n\nChoose the right configuration\n------------------------------\n\nThis section describes the values that you can set for the configuration.\nChanging these values affects the recommendations that you receive.\n\n### The observation period\n\nSet the observation period duration to calculate recommendations by modifying\nthe value for `idle_reservation_lookback_period` or\n`under_utilized_reservation_lookback_period` and upload the new\nconfiguration for your project. You can set the observation period\nto a value between 7 days and 30 days, for example:\n\n- For an observation period of the previous 7 days, use `\"P7D\"`.\n- For an observation period of the previous 30 days, use `\"P30D\"`.\n\nBy default, the observation period is 7 days.\n\n- For recommendations based on short-term changes in your workload, use a shorter observation period.\n- For recommendations that are not affected by short-term fluctuations in your workload, use a longer observation period.\n\nSimilarly, set the usage threshold that triggers an underutilized reservation\nrecommendation by modifying the value for\n`under_utilized_reservation_utilization_threshold` and upload the new\nconfiguration for your project, for example:\n\n- For a threshold of 80%, `\"0.8\"`.\n\nWhat's next\n-----------\n\n- Learn how to [view and apply idle reservation recommendations](/compute/docs/instances/view-and-apply-idle-reservation-recommendations) or [view and apply underutilized reservation recommendations](/compute/docs/instances/view-and-apply-underutilized-reservation-recommendations).\n- Learn how to [configure idle reservation recommendations](/compute/docs/instances/configure-idle-reservation-recommendations) or [configure underutilized reservation recommendations](/compute/docs/instances/configure-underutilized-reservation-recommendations)."]]