[[["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):2025-09-03。"],[[["\u003cp\u003eRecommender helps minimize cold starts in Google Cloud functions by detecting functions that would benefit from setting minimum instances.\u003c/p\u003e\n"],["\u003cp\u003eRecommendations are automatically provided for functions one week after deployment and are displayed in the Cloud Run functions Overview page.\u003c/p\u003e\n"],["\u003cp\u003eUsers can view recommendations, get details on why they're suggested, and directly apply the recommendation to set minimum instances.\u003c/p\u003e\n"],["\u003cp\u003eIn addition to the Cloud Run functions UI, recommendations are accessible through the Recommender API, BigQuery export, and Recommender Logs.\u003c/p\u003e\n"],["\u003cp\u003eRecommendations can be dismissed for 30 days if they are not applicable, and recommendations can also be found in the Recommendation Hub.\u003c/p\u003e\n"]]],[],null,["# Minimize Cold Starts with Recommender\n=====================================\n\nRecommender is a service that\nautomatically provides recommendations and insights for using resources on\nGoogle Cloud, based on heuristic methods, machine learning, and current\nresource usage. Each recommendation includes a link you can click to put the\nrecommendation into effect for your function.\n\nThis topic shows how to use [Recommender](/recommender) to minimize\ncold starts:\n\n- Detect functions that could benefit from setting up the minimum instances (in other words, functions that are likely to be negatively affected by cold start issues).\n- Inform users about the opportunity to eliminate cold start issues.\n- Let users set up the minimum instances directly from the recommendation system.\n\nRecommendation availability after deployment\n--------------------------------------------\n\nRecommender automatically provides recommendations for a function as\nsoon as a week after it has been deployed.\nAfter this period of time, recommendations for the function are displayed with\nthe function on the function list page in the\n[Cloud Run functions Overview page](https://console.cloud.google.com/functions/list)\nin the Google Cloud console and in the\n[Recommendation Hub](/recommender/docs/recommendation-hub/identify-configuration-problems).\n\nAlternate ways of using recommendations\n---------------------------------------\n\nIn addition to the use of recommendations covered on this page inside the\nCloud Run functions UI, recommendations are also available through the following:\n\n- [Recommender API](/recommender/docs/using-api)\n- [BigQuery export](/recommender/docs/bq-export/export-recommendations-to-bq)\n- [Recommender Logs](/recommender/docs/viewing-logs)\n\nViewing and accepting recommendations for Cloud Run functions\n-------------------------------------------------------------\n\nTo view and accept a recommendation in the Cloud Run functions user interface:\n\n1. [Go to the Cloud Run functions Overview page](https://console.cloud.google.com/functions/list)\n\n2. Look for functions in the list that have a **Performance** light bulb\n icon in the **Recommendation** column.\n\n3. Click **Performance** to view details for the recommendation:\n\n - An overview of why this recommendation is being suggested.\n - The recommendation itself.\n4. If you decide to apply the recommendation, set the number of minimum\n instances and click **Apply**.\n\n This redeploys the function with the new minimum instance value. After this,\n the recommendation will no longer be displayed next to the function.\n\nDismissing a recommendation\n---------------------------\n\nClick **Dismiss** if you want to dismiss the recommendation without applying it.\nThis prevents the recommendation for that function from appearing again for\n30 days.\n\nViewing recommendations in Recommendation Hub\n---------------------------------------------\n\nYou can also view recommendations in the Recommendation Hub.\nFor more information refer to the Recommendation Hub\n[Getting started](/recommender/docs/recommendation-hub/identify-configuration-problems)\ndocumentation."]]