亞馬遜re:Invent10週年官宣重大發布，推出自研芯片Graviton3、5G、IOT、數字孿生等多項服務

{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"11月29日，一年一度的亞馬遜雲科技re:Invent全球大會開幕，今年也是re:Invent十週年。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"re:Invent 2021大會首日，亞馬遜雲科技發佈了涵蓋計算、物聯網、5G、無服務器數據分析、大機遷移、機器學習等方向的多項新服務和功能："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"包括髮布基於新一代自研芯片Amazon Graviton3的計算實例、幫助大機客戶向雲遷移的Amazon Mainframe Modernization、幫助企業構建移動專網的Amazon Private 5G、四個亞馬遜雲科技分析服務套件的無服務器和按需選項以及爲垂直行業構建的雲服務和解決方案，如構建數字孿生的服務Amazon IoT TwinMaker和幫助汽車廠商構建車聯網平臺的Amazon IoT FleetWise等。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"新任亞馬遜雲科技首席執行官Adam Selipsky首次亮相re:Invent全球大會。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"演講中， Adam Selipsky首先回顧了亞馬遜雲科技發展過程中的重要里程碑。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"2006年，"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#333333","name":"user"}}],"text":"Amazon Simple Storage Service (Amazon S3) 簡單存儲"},{"type":"text","text":"問世。時至今日，Amazon S3已成爲業界公認的最重要的數據存儲標準之一，存儲超過100萬億個對象文件。"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" Amazon S3推出幾個月後，Amazon Elastic Cloud Compute (Amazon EC2)彈性計算問世，隨即推出第一款數據庫服務，從此雲具備了有能力承載應用程序的所有基礎元素。"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" 如今，亞馬遜雲科技擁有475+不同實例類型，通用型、計算優化型、內存優化型、存儲優化型、硬件加速型......幾乎適用所有工作負載。每天有超過6000萬個新Amazon EC2實例啓動。"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"亞馬遜雲科技現提供超過200項全功能的服務，涵蓋計算、存儲、數據庫、數據分析、機器學習與人工智能...."}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Adam表示，過去十五年，雲計算從被質疑到成爲新一代 IT 標準，並發展成影響整個 ICT 產業的創新和革命。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"不過在Adam看來，雲計算還處在非常早的時期。他援引分析師的估計，目前雲上支出只佔整個IT支出的5%到15%。未來將有大量的工作負載遷移到雲，大量的創新即將在雲上開展，雲產業前景巨大。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"在亞馬遜雲科技看來，儘管雲計算已經極大地提高了IT的效率、敏捷性、彈性，但依然可以探索新的方式，讓雲變得更加靈活、成本更優、延遲更低。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"發佈三款由自研芯片支持的新Amazon EC2實例"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Adam Selipsky 表示，“儘管我們已經如此努力地創新，我們依然意識到，如果希望針對所有可能的工作負載徹底變革計算的性價比，我們需要徹底重新思考實例。爲了實現這個目的，我們需要深入底層技術，一直到芯片。因此，我們開始自己設計基於Arm的芯片。”"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"},{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"在"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"re:Invent大會現場，亞馬遜雲科技宣佈推出三款由自研芯片支持的新Amazon Elastic Compute Cloud (Amazon EC2)實例，幫助客戶提升在Amazon EC2上運行的工作負載的性能、成本和能源效率。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}},{"type":"strong"}],"text":"新C7g 實例由Amazon Graviton3 處理器支持，與由 Amazon Graviton2 處理器支持的當前一代 C6g 實例相比，可將計算密集型工作負載性能提高多達25%。"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Amazon Graviton3處理器與Graviton2相比，爲科學計算、機器學習和媒體編碼工作負載提供高達2倍的浮點運算性能，爲加密工作負載速度提升高達2倍，爲機器學習工作負載提供高達3倍的性能。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Amazon Graviton3處理器的能效也更高，在相同性能下，與同類型EC2實例對比，可節省高達60%的能源消耗。C7g實例是雲中第一個採用最新DDR5內存的實例，與基於Amazon Graviton2的實例相比，它提高了50%的內存帶寬，從而提高了科學計算等內存密集型應用的性能。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"與基於Amazon Graviton2的實例相比，C7g實例的網絡帶寬也高出20%。C7g 實例支持 Elastic Fabric Adapter (EFA)，允許應用程序直接與網絡接口卡通信，提供更低且更一致的延遲，提高需要大規模並行處理（如 HPC 和視頻編碼）的應用程序的性能。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"C7g實例現已提供預覽版。欲瞭解更多C7g實例的信息，請訪問："},{"type":"link","attrs":{"href":"https:\/\/aws.amazon.com\/ec2\/instance-types\/c7g\/","title":null,"type":null},"content":[{"type":"text","marks":[{"type":"underline"}],"text":"aws.amazon.com\/ec2\/instance-types\/c7g"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}},{"type":"strong"}],"text":"由Amazon Trainium芯片支持的新Trn1實例爲在Amazon EC2中訓練絕大多數機器學習模型提供最佳性價比，及最快的訓練速度。"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"越來越多客戶正在構建、訓練和部署機器學習模型。爲了確保提高準確性，這些機器學習模型必須構建在越來越多的訓練數據上，導致其訓練成本越來越高。這可能會限制客戶能夠部署的機器學習模型數量。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"亞馬遜雲科技爲機器學習提供了更多計算服務選擇，包括採用NVIDIA A100 Tensor Core GPU的EC2 P4d實例和採用Habana Labs  Gaudi 加速器的 EC2 DL1 實例。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"但即使擁有當今最快的加速實例，訓練持續變大的機器學習模型仍然是非常昂貴和耗時的。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Trn1實例爲在Amazon EC2中進行深度學習模型訓練提供最佳性價比以及最快的訓練速度，與P4d實例相比，通過Trn1實例訓練深度學習模型的成本降低多達40%。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Trn1實例提供800Gbps EFA網絡帶寬（比最新基於GPU的EC2實例高兩倍），並與Amazon FSx for Lustre高性能存儲集成，讓客戶可以啓動具有EC2 UltraClusters功能的Trn1實例。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"通過EC2 UltraClusters，開發人員可以將機器學習訓練擴展到一萬多個與 PB 級網絡互連的 Trainium 加速器，讓客戶按需訪問超算級性能，即便是最大型和最複雜的模型，訓練時間也可以從幾個月縮短到幾天。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Trn1實例現已提供預覽版。欲瞭解更多Trn1實例的信息，請訪問"},{"type":"link","attrs":{"href":"https:\/\/aws.amazon.com\/ec2\/instance-types\/trn1","title":null,"type":null},"content":[{"type":"text","marks":[{"type":"underline"}],"text":"aws.amazon.com\/ec2\/instance-types\/trn1"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}},{"type":"strong"}],"text":"基於自研Amazon Nitro SSDs（固態驅動器）的新存儲優化型Im4gn\/Is4gen\/ I4i實例爲在Amazon EC2上運行的I\/O密集型工作負載提供最佳存儲性能。"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Im4gn\/Is4gen\/I4i實例旨在通過架構最大限度提高I\/O密集型工作負載的存儲性能。通過自研的 Amazon Nitro SSDs ，Im4gn\/Is4gen\/I4i實例提供高達 30 TB 的 NVMe 存儲，與上一代I3實例相比，I\/O 延遲降低了 60%，延遲可變性降低了 75%，從而最大限度地提高了應用程序性能。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Amazon Nitro SSDs通過優化存儲堆棧、虛擬化管理程序和硬件與Amazon Nitro 系統緊密集成。與使用商用SSD相比，亞馬遜雲科技同時管理Amazon Nitro SSDs的硬件和固件，使SSD更新交付速度更快，讓客戶可以從改進的功能中獲益。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Im4gn 實例（現已可用）採用 Amazon Graviton2 處理器，與 I3 實例相比，性價比提高多達 40%，每 TB存儲成本降低多達 44%。Is4gen 實例（現已可用）也採用 Amazon Graviton2 處理器，與 I3en 實例相比，每 TB 存儲成本降低多達 15%，計算性能提高多達 48%。開始使用Im4gn\/Is4gn實例，請訪問"},{"type":"link","attrs":{"href":"http:\/\/aws.amazon.com\/ec2\/instance-types\/i4g","title":null,"type":null},"content":[{"type":"text","marks":[{"type":"underline"}],"text":"aws.amazon.com\/ec2\/instance-types\/i4g"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"I4i實例（即將可用）採用英特爾第三代可擴展處理器（Ice Lake），與當前一代 I3 實例相比，計算性能提高多達 55%。欲瞭解更多Im4gn\/Is4gen\/ I4i實例的信息，請訪問"},{"type":"link","attrs":{"href":"https:\/\/aws.amazon.com\/ec2\/instance-types\/i4i","title":null,"type":null},"content":[{"type":"text","marks":[{"type":"underline"}],"text":"aws.amazon.com\/ec2\/instance-types\/i4i"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"發佈雲上全面數據解決方案"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"數據是企業洞察的驅動器。亞馬遜雲科技在雲上推出了全面的數據解決方案，從數據庫、到數據分析，到人工智能，希望幫企業獲得數據洞察，更好地推動業務發展。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Adam在主題演講中宣佈，亞馬遜雲科技分析服務套件"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}},{"type":"strong"}],"text":"新推出三種無服務器選項和一個按需選項"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Amazon Redshift Serverless可在幾秒鐘內自動設置和擴展資源，讓客戶在PB級數據規模運行高性能工作負載，而無需管理數據倉庫集羣；"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Amazon Managed Streaming for Apache Kafka (Amazon MSK) Serverless可快速擴展資源，極大地簡化實時數據攝取和流式傳輸；"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Amazon EMR Serverless讓客戶無需部署、管理和擴展底層基礎設施，即可使用開源大數據框架（如 Apache Spark、Hive 和 Presto）運行分析應用程序。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Amazon Kinesis On-demand是亞馬遜雲科技爲Amazon Kinesis Data Streams（一種用於捕獲、處理和存儲流媒體數據的無服務器服務）推出的新功能（新的容量模式），利用它可以提供每分鐘GB級別的讀寫吞吐量，而無需進行容量規劃。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"機器學習是數據洞察必不可少的。亞馬遜雲科技擁有廣泛和完整的機器學習功能特性組合，從機器學習框架和基礎設施、機器學習服務到開箱即用的人工智能服務。亞馬遜雲科技的機器學習服務Amazon SageMaker，目前有數萬客戶正在使用，訓練具有數十億個參數的模型，每月進行數千億次預測。自2017年問世以來，Amazon SageMaker已經陸續添加了150多項功能和特性。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"此外，Adam還"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}},{"type":"strong"}],"text":"宣佈了一項重要的Amazon SageMaker新功能——Amazon SageMaker Canvas"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"，它讓業務人員和數據分析人員能夠使用可視化的點選界面，生成高準確度的機器學習預測，不需要寫程序代碼。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"亞馬遜雲科技還爲其數據湖服務"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}},{"type":"strong"}],"text":"Amazon Lake Formation增加了兩個功能"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"，Row and cell-level security for Lake Formation爲湖中數據提供精確到數據行和單元格的細顆粒度訪問控制；Transactions for governed tables in Lake Formation確保湖中數據的一致性，這兩項功能讓數據湖的搭建、治理和管理更加容易。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"推出兩項IoT新服務，賦能行業"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"大會現場，亞馬遜雲科技還"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}},{"type":"strong"}],"text":"推出了兩項IoT新服務—— Amazon IoT TwinMake\/FleetWise"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"，賦能更廣泛的行業。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"中橋調研諮詢（Sino-bridges）首席分析師王叢認爲：“Amazon IoT TwinMaker打通物理和虛擬世界，讓物理世界監控管理更加實時，高效，智能。另外，Amazon IoT FleetWise讓這種數實融合的綜合管理能力轉化爲車隊物聯網監控、分析、管理創新服務。”"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"Amazon IoT TwinMaker 可更輕鬆、快捷"},{"type":"text","marks":[{"type":"italic"}],"text":"創建現實世界的數字孿生"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}},{"type":"strong"}],"text":"Amazon IoT TwinMaker"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"讓開發人員更加輕鬆、快捷地創建現實世界的數字孿生，如樓宇、工廠、工業設備和生產線。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}},{"type":"strong"}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"數字孿生是物理系統的虛擬映射，可根據其所代表的現實世界對象的結構、狀態和行爲定期更新。Amazon IoT TwinMaker讓開發人員可以輕鬆彙集來自多個來源（如設備傳感器、攝像機和業務應用程序）的數據，並將這些數據結合起來創建一個知識圖譜，對現實世界環境進行建模。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"客戶可以通過Amazon IoT TwinMaker，使用數字孿生來構建反映現實世界的應用程序，提高運營效率並減少停機時間。使用Amazon IoT TwinMaker無需預付費用，客戶只需爲使用的服務付費。使用Amazon IoT TwinMaker，請訪問："},{"type":"link","attrs":{"href":"https:\/\/aws.amazon.com\/cn\/iot-twinmaker\/","title":null,"type":null},"content":[{"type":"text","marks":[{"type":"underline"}],"text":"aws.com\/iot-twinmaker"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"製造企業從設備傳感器、攝像機和業務應用程序（例如企業資源規劃系統或項目管理系統）等來源收集和處理有關其設備和設施的大量數據。許多客戶希望結合這些數據源來創建其物理系統的虛擬映射（稱爲數字孿生），幫助他們模擬和優化運營績效。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"但是，即便對技術最先進的企業來說，構建和管理數字孿生也很困難。爲了構建數字孿生，客戶必須手動連接來自不同來源的不同類型的數據（例如來自設備的時間序列傳感器數據、來自攝像機的視頻傳送、來自業務應用程序的維護記錄等）。然後，客戶必須創建一個知識圖譜，提供對所有連接數據的通用訪問，並將數據源之間的關係映射到物理環境。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"爲了實現數字孿生，客戶必須構建其物理系統（例如樓宇、工廠、設備、生產線等）的 3D 虛擬模型，並將現實世界的數據疊加到 3D 視圖上。一旦他們獲得具有實時數據的現實世界的虛擬映射，客戶就可以爲工廠操作員和維護工程師構建應用程序，並利用機器學習和數據分析來洞察其物理系統的實時操作性能。由於上述所需工作的複雜性，絕大多數企業無法使用數字孿生來改善其運營。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Amazon IoT TwinMaker讓創建現實世界的數字孿生變得更加輕鬆、快捷。使用Amazon IoT TwinMaker，開發人員可以將Amazon IoT TwinMaker連接到設備傳感器、視頻源和業務應用程序等數據源，快速開始構建設備、裝置和流程的數字孿生。爲方便從各種數據源收集數據，Amazon IoT TwinMaker包含適用於Amazon IoT SiteWise、Amazon Kinesis Video Streams和Amazon S3的內置連接器（客戶也可以爲Amazon Timestream或Snowflake等數據源添加自己的連接器）。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Amazon IoT TwinMaker會自動創建一個知識圖譜，整合並理解所連接數據源的關係，因此它可以使用被映射系統的實時信息更新數字孿生。客戶可以將現有的3D模型（例如CAD和BIM文件、點雲掃描等）直接導入Amazon IoT TwinMaker，輕鬆創建物理系統（例如樓宇、工廠、設備、生產線等）的3D視圖，並將知識圖譜中的數據疊加到3D視圖上，創建數字孿生。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"數字孿生創建完畢後，開發人員就可以使用適用於Amazon Managed Grafana的Amazon IoT TwinMaker插件創建基於Web的應用程序，在工廠操作員和維護工程師用於監控和檢查設施和工業系統的設備上，即可顯示該應用程序的數字孿生。通過Amazon IoT TwinMaker，更多客戶可以使用數字孿生模型構建模擬現實世界的應用程序，提高運營效率並減少停機時間。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"據悉，Amazon IoT TwinMaker現已在美國東部（弗吉尼亞北部）、美國西部（俄勒岡）、亞太地區（新加坡）和歐洲（愛爾蘭）區域提供預覽，其他區域也將很快推出。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"在線下媒體會上，對於目前火熱的元宇宙概念，亞馬遜全球副總裁、亞馬遜雲科技大中華區執行董事張文翊對InfoQ等媒體表示，元宇宙是雲計算可以大量賦能的一個領域。她表示：“我們認爲元宇宙一定是雲計算可以大量賦能的一個領域。元宇宙本身需要的就是計算、存儲、機器學習等，這些都離不開雲計算。比如Epic Games是個開發元宇宙遊戲Fortnite（堡壘之夜）的公司，全球擁有3.5億用戶，它的工作負載幾乎全部都跑在亞馬遜雲科技上。”"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"亞馬遜雲科技大中華區產品部總經理顧凡如此看待火熱的元宇宙概念：“元宇宙裏面融合了大家所熟知的大量技術，而這些技術的背後就是雲計算。無論是在數字世界裏面建模，還是與之互動，都離不開機器學習在背後支撐。與此同時，元宇宙的世界必然會產生超級大的數據量，這些數據都離不開底層平臺，而云一定是最佳的平臺。”"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"Amazon IoT FleetWise幫助汽車製造商收集、管理車輛數據"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"亞馬遜雲科技還宣佈推出了Amazon IoT FleetWise，讓汽車製造商更輕鬆、經濟地收集、管理車輛數據，同時幾乎實時上傳到雲端。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"通過Amazon IoT FleetWise，汽車製造商可以輕鬆地收集和管理汽車中任何格式的數據（無論品牌、車型或配置），並將數據格式標準化，方便在雲上輕鬆進行數據分析。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"Amazon IoT FleetWise的智能過濾功能，幫助汽車製造商近乎實時地將數據高效上傳到雲端，爲減少網絡流量的使用，該功能也允許開發人員選擇需要上傳的數據，還可以根據天氣條件、位置或汽車類型等參數來制定上傳數據的時間規則。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"當數據進入雲端後，汽車製造商就可以將數據應用於車輛的遠程診斷程序，分析車隊的健康狀況，幫助汽車製造商預防潛在的召回或安全問題，或通過數據分析和機器學習來改進自動駕駛和高級輔助駕駛等技術。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"使用Amazon IoT FleetWise無需預付費用，客戶只需爲使用的基礎服務付費。欲瞭解Amazon IoT FleetWise，請訪問"},{"type":"link","attrs":{"href":"https:\/\/aws.amazon.com\/iot\/kaleidoscope","title":null,"type":null},"content":[{"type":"text","marks":[{"type":"underline"}],"text":"aws.amazon.com\/iot\/fleetwise"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"具有標準車輛數據收集功能的Amazon IoT FleetWise現已在美國東部（弗吉尼亞北部）和歐洲（法蘭克福）區域提供預覽，其他區域也將很快推出。Amazon IoT FleetWise先進的攝像頭傳感器數據收集功能預計將於2022年上半年推出。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"新產品和新服務何時落地中國？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"提及此次發佈的產品和服務何時落地中國，產品和服務面向國內市場和美國市場有何不同時，"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}},{"type":"strong"}],"text":"亞馬遜雲科技大中華區產品部總經理顧凡對InfoQ等媒體"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"表示：“我們有的時候不能直接簡單去看這個產品到底中外有什麼差異，而是要看我們是否已經把中國客戶的需求結合到產品設計裏面。比如今天提到的Graviton2、Graviton3，我們一定會加速把Graviton3帶到中國。再比如，全球SageMaker同步首發的六個區域，中國就佔了兩個，就是因爲中國客戶的機器學習場景非常多。”"}]}]}