---
格式版本: 2
标题: "Amazon ECS introduces Early Success Criteria for service deployments"
原文链接: "https://aws.amazon.com/cn/about-aws/whats-new/2026/09/amazon-ecs-deployments-early-success/"
发布日期: "2026-09-04"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "llm:local:original_script_field"
发布时间证据: "postDateTime: 2026-09-04T17:00:00Z"
发布时间校准原因: "候选日期来自文章内的postDateTime字段，明确标注为发布日期，且符合AWS What's New公告格式。"
发布时间校准置信度: "1"
发布时间候选数量: 1
发布时间严格候选数量: 0
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-09-07T22:33:23+08:00"
发布时间仲裁状态: "confirmed"
发布时间仲裁尝试次数: 1
发布时间仲裁耗时毫秒: 4576
发现时间: "2026-09-07T22:07:26+08:00"
入库时间: "2026-09-07T14:33:34.960Z"
来源平台: "固定入口"
搜索渠道: "fixed_url"
搜索词: "https://aws.amazon.com/new"
匹配关键词:
  - "deployment"
  - "GPU"
相关厂家:
  - "AWS"
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 39
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "正文主线为Amazon ECS部署功能Early Success Criteria，属于容器服务部署生命周期/运维能力，非超节点、AI Rack或机架级基础设施；仅提及GPU加速推理为适用场景，主题相关性低。来源为AWS官方what's new，权威性高但主题错位。知识库无该功能历史记录，本文对超节点情报库无新增可核验机架级事实。虽有healthy percent、BLOCKING/DEFERRED清理等部署机制细节，但属于单厂商容器服务运维功能，不构成可复用机架级基础设施机制。命中‘教程与运维选型’否决项，无新机架硬件/协议标准、生产级基准或规模部署里程碑。"
AI质检模型: "zj-deepseek-v4-flash"
AI质检时间: "2026-09-07T22:33:59+08:00"
AI主题相关性: 3
AI来源权威性: 13
AI新颖性: 6
AI技术细节: 5
AI商业部署信号: 3
AI完整性: 9
AI评分提示词版本: "v17-精简生产版"
AI评分提示词SHA256: "48fb9777f386026761b4873eaff30807694fb11e9b352d7c69bf2dfde750cc7d"
AI评分知识库版本: "knowledge_base_v1-20260819+runtime.97"
AI评分知识库SHA256: "093c22b302b0d377ed12eab9c169d422553a099393d75a2ea63f76a9eb1d8782"
AI评分知识库检索词: "[\"AWS\",\"deployment\",\"https://aws.amazon.com/new\",\"RAS\",\"GPU\",\"ECS\",\"CI/CD\",\"GPU-accelerated\",\"BLOCKING\",\"DEFERRED\",\"US\",\"CLI\"]"
AI评分知识库命中: "[{\"id\":\"runtime-b9ec799d769b0f4ba49df311\",\"title\":\"AWS and NVIDIA to Deliver 2 Million Additional GPUs and Next-Generation Infrastructure for Agentic and Physical AI\",\"sourceType\":\"ai_excellent_article\",\"time\":\"\",\"matchedTerms\":[\"AWS\",\"deployment\",\"RAS\",\"GPU\",\"GPU-accelerated\",\"US\"],\"rank\":-14.245043415344622},{\"id\":\"runtime-9f43a79e9c23b1f329d8a71a\",\"title\":\"AWS and NVIDIA to Deliver 2 Million Additional GPUs and Next-Generation Infrastructure for Agentic and Physical AI\",\"sourceType\":\"ai_excellent_article\",\"time\":\"\",\"matchedTerms\":[\"AWS\",\"deployment\",\"RAS\",\"GPU\",\"GPU-accelerated\",\"US\"],\"rank\":-13.497944746359234},{\"id\":\"runtime-676f90ee5f88ebab3b504077\",\"title\":\"NVIDIA NVLink Fusion Brings NVHBM to Next-Generation AI Infrastructure\",\"sourceType\":\"ai_excellent_article\",\"time\":\"\",\"matchedTerms\":[\"AWS\",\"deployment\",\"RAS\",\"GPU\",\"GPU-accelerated\",\"US\"],\"rank\":-11.440723443450551},{\"id\":\"runtime-15c79e6868a9b43f99c061a9\",\"title\":\"‘Out of Hyperbole’: Nvidia’s AI Boom Tests Data Center Infrastructure Limits\",\"sourceType\":\"ai_excellent_article\",\"time\":\"2026-08-27\",\"matchedTerms\":[\"AWS\",\"deployment\",\"RAS\",\"GPU\",\"GPU-accelerated\",\"US\",\"CLI\"],\"rank\":-11.108832385191004},{\"id\":\"runtime-bf4039a0342d37545e9459a2\",\"title\":\"Most Neoclouds Suck At Security\",\"sourceType\":\"ai_excellent_article\",\"time\":\"2026-08-30\",\"matchedTerms\":[\"AWS\",\"deployment\",\"RAS\",\"GPU\",\"US\",\"CLI\"],\"rank\":-9.582679476801044}]"
采集批次: "2026年9月7日22点05分18秒"
采集批次ID: "20260907-220517-361"
去重键: "https://aws.amazon.com/cn/about-aws/whats-new/2026/09/amazon-ecs-deployments-early-success"
---

[Amazon Elastic Container Service](https://aws.amazon.com/ecs/) (Amazon ECS) now supports Early Success Criteria for rolling service deployments, giving you the flexibility to define when a deployment is considered successful based on your confidence level and the operational needs of your workload. This can help you complete deployments sooner and unblock subsequent deployments, CI/CD pipelines, and other dependent operations.

With Early Success Criteria, you configure the healthy percent - the proportion of desired tasks that must be running and healthy on the target service revision before the deployment is marked successful. For example, with a desired count of 100 and a healthy percent of 90%, Amazon ECS marks the deployment successful after 90 tasks are healthy and continues launching the remaining tasks through regular service scaling, outside the deployment lifecycle. This can benefit workloads running on specialized or constrained capacity, such as GPU-accelerated inference workloads, where hardware availability can extend task launch times. Early Success Criteria also gives you more control over how long deployment rollback monitoring applies, allowing it to protect the deployment until your configured success criteria are met while subsequent scale-out continues through regular service scaling. You can also choose how Amazon ECS handles source service revision cleanup using BLOCKING or DEFERRED. With BLOCKING, Amazon ECS completes source revision cleanup before declaring success. With DEFERRED, Amazon ECS declares success when the criteria are met and drains source revision tasks asynchronously outside the deployment. This benefits services with active long-lived connections or task scale-in protection, where source revision tasks may need to remain running without holding the deployment open.

The feature is available with the rolling deployment strategy in all AWS Commercial and AWS GovCloud (US) [Regions](https://aws.amazon.com/about-aws/global-infrastructure/regional-product-services/). You can configure Early Success Criteria for new and existing Amazon ECS services using the AWS Management Console, AWS CLI, AWS SDKs, and infrastructure as code (IaC) tools. To learn more, see our [documentation](https://docs.aws.amazon.com/AmazonECS/latest/developerguide/early-success-criteria.html).