洞察數據庫變革趨勢，亞馬遜雲科技正在憑藉這項技術改變着遊戲規則

{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"如果要對遺留的數據庫進行升級，沒有比上雲更好的解決方案了。合適的雲平臺可以提供可用性強、伸縮性好、安全性高的商業數據庫，而不需要高昂的價格。但這其中也存在着一定的風險——糟糕的遷移可能會使關鍵型工作負載和敏感客戶數據處於危險之中，沒有犯錯的餘地。"}]},{"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":"幸運的是，亞馬遜雲科技提供了大量的工具來幫助組織遷移和更新雲上的數據庫。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"數據庫技術和生態正在經歷一場變革"}]},{"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":"互聯網、移動互聯網應用開發浪潮的崛起，產生了大量的NoSQL數據，面對電商、視頻、社交、出行類場景時，無論是從客戶併發角度，還是從要處理數據級別角度，對數據庫的要求之高，這並不是一個傳統的關係型數據庫就能處理的，這也驅動了雲原生關係型數據庫的出現。"}]},{"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":"在這樣的背景下，越來越多的用戶希望把應用架構在開源數據庫引擎上，無論是MySQL還是PostgreSQL，而不願意再把自己的應用架構在一個有嚴苛license的、會被鎖定的數據庫引擎上。此外，現代化應用也在不斷被提起，擁抱現代化應用的客戶都希望自己的企業更加敏捷，希望讓開發更專注於應用和數據，向着全託管式數據庫遷移。"}]},{"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":"軟件架構的迭代和轉型也驅動着數據庫選型的變化以及數據庫技術和架構的迭代。過去大家做軟件程序開發，從早年的PC、互聯網、移動互聯網，到現在的萬物互聯，軟件架構基本上都是從Server到SOA再到微服務這樣一個軟件架構的迭代和轉型，當然它也在驅動着數據庫選型的變化。"}]},{"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":"無疑，數據庫的技術和生態勢必要經歷一場深刻的變革。面對這種趨勢，亞馬遜雲科技早已做好了應對之策。"}]},{"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表示，亞馬遜雲科技在做雲上數據庫產品時有五大理念：第一是專庫專用，極致性能；第二是無服務器，敏捷創新；第三是全球架構，一鍵部署；第四是平滑遷移，加速上雲；第五是AI賦能，深度集成。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"專庫專用，極致性能"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"在數據爆炸式增長，微服務架構與DevOps愈發流行的今天，一個數據庫打天下的時代早就過去了。市場上真正需要的是爲開發者在不同的應用場景下針對不同的數據類型和不同的數據訪問特點，提供專門構建的工具。亞馬遜雲科技目前提供十多種專門構建的數據庫服務，支持關係、鍵值、文檔、內存、圖、時間序列、寬列和分類賬八大數據類型。"}]},{"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":"目前，亞馬遜雲科技專門構建數據庫的產品家族主要包括：託管的關係型數據庫Amazon RDS（它開啓了託管數據庫服務的新模式，從最初只支持MySQL，到目前支持多種常用的數據庫引擎，包括MySQL、PostgreSQL、Oracle、SQL Server、MariaDB）；開源的數據庫Amazon Aurora；非關係型數據庫Amazon DynamoDB以及圖數據庫Amazon Neptune等。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"無服務器，敏捷創新"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"無服務器、敏捷創新。其實無服務器被提及了很久，但是我們很多人不理解爲什麼需要無服務器。絕大多數企業在做數據庫規劃時，都是按照峯值留有餘地的方式計算資源，這勢必會帶來成本的浪費，而無服務器最大的優勢就是可以讓數據庫的管理和運維解放。"}]},{"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":"今天，我們已經可以做到手動或者自由地去調整數據庫計算資源，但是這背後如果要真的實現在生產上面根據業務需求去擴展數據庫規模，背後還需要非常多的技術上的努力，才能夠真正將它實現。"}]},{"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":"如果從運維的層面來解讀，無服務器模式背後所代表的和雲計算的概念是非常相近的。也就是說，將雲廠商這麼多年的對於大規模數據庫運維、伸縮擴展的經驗，和團隊整個的技術能力最大化開放給客戶，使客戶受益。希望通過亞馬遜雲科技自己的業務去利用到Amazon  DynamoDB這樣一種Serverless，讓客戶真正受益。"}]},{"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":"Amazon DynamoDB一直使用無服務器的模式對外提供服務；Amazon Aurora Serverless V2可以在幾分之一秒內將數據庫工作負載從數百個事務擴展到數十萬個事務，與按照峯值負載配置容量的成本相比，最多可節省 90% 的數據庫成本。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"全球架構，一鍵部署"}]},{"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":"在全球化的今天，無論是手機、視頻娛樂還是現在非常流行的智能汽車，我們的客戶羣體往往分佈到全球，疫情後的恢復工作對於我們現代化的、全球化的分佈式應用架構來說也提出了新的挑戰。因此，支持全球業務的這些應用，包括數據庫，需要保證全球業務的連續性，爲全球客戶提供一致的、低延遲的用戶體驗。"}]},{"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":"亞馬遜雲科技提供了具備全球數據庫能力的數據庫服務，包括Amazon Aurora  Global Database、Amazon DynamoDB  Global Tables、Amazon ElastiCache for Redis Global Datastore、Amazon DocumentDB Global Clusters。都能夠利用亞馬遜雲科技的骨幹網絡提供比互聯網更穩定的網絡支撐，以一鍵部署的方式，幫助客戶實現幾千公里跨區域數據庫災備，故障恢復大概能在一分鐘之內完成，同時跨區域的數據複製延遲通常小於一秒。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"平滑遷移，加速上雲"}]},{"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":"目前，全球已經有超過45萬的數據庫通過亞馬遜雲科技的數據庫遷移服務實現了遷移到亞馬遜雲科技雲平臺。從工具角度上，亞馬遜科技希望客戶能夠通過Amazon DMS、Amazon Database Migration Service，進行自助式的雲遷移。對於遷移過程中可能會需要的人才以及相應的規劃，包括遷移之後的優化、相應的流程，亞馬遜雲科技會通過專業的服務團隊和合作夥伴網絡成員，爲客戶提供專家方面的支持。"}]},{"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":"服務包括Amazon Schema Conversion Tool可轉換數據庫架構，Amazon Database Migration Service可遷移數據，新發布的Babelfish for Amazon Aurora PostgreSQL則可幫助客戶遷移SQL Server應用程序。Babelfish for PostgreSQL源代碼現已在Github開源。客戶還可以通過亞馬遜雲科技Database Freedom項目，瞭解如何從傳統商業數據遷移至雲數據庫。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"AI賦能，深度集成"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"構建現代化應用尤其是用數據驅動去做創新時，已經無數次的證明，使用機器學習的方式要比傳統方式更好。但將數據庫與機器學習相結合也會面臨一些挑戰。亞馬遜雲科技的解決方案是——Amazon Neptune。"}]},{"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":"Neptune是專門構建的針對於圖的數據結構，能夠存儲物品、人物、各種事物之間的關係的數據庫，在Amazon Neptune裏面也有中國區域可用的Neptune ML的功能，允許數據工程師不需要掌握機器學習的技能，直接從圖數據庫裏導出數據、轉換格式、訓練模型併發布。Amazon Neptune是數據庫與AI的深度集成，數據庫開發者無需具備機器學習專業知識，只需使用熟悉的數據庫語言，即可進行機器學習操作。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"Amazon Aurora的三大優勢"}]},{"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":"多年來，亞馬遜雲科技不斷加速數據庫相關服務落地，並針對不同應用場景需求推出了多種數據庫服務產品，爲客戶提供兼具高性能、高可用性、可擴展及成本效益的專用數據庫服務，支撐客戶現代化應用的快速部署及創新。"}]},{"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":"既然亞馬遜已經有了很多像Amazon RDS這樣的服務產品了，爲什麼還要推出Amazon Aurora？"}]},{"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":"對於這個問題，顧凡表示："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"客戶的聲音非常簡單，要的就是MySQL、PostgreSQL這樣的開源數據庫引擎，但是我們去看開源版本的MySQL和PostgreSQL是達不到企業級商業數據庫所達到的性能，也達不到企業級商業數據庫的可用性。所以客戶要的是魚和熊掌兼得，兩者都要，既要開源的數據庫引擎，又要跟商業數據庫一樣的可用性。而Amazon Aurora的目的就是既要像使用MySQL、PostgreSQL這樣的開放數據庫引擎一樣節省成本，同時也要能夠提供商業級數據庫那樣的性能、持久性和可用性。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Amazon Aurora可以提供5倍於標準MySQL的性能，以及3倍於PostgreSQL的吞吐量。同時提供高可用，可以擴展到15個只讀副本，同時成本只有商業數據庫的1\/10。"}]}]},{"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":"據顧凡介紹，Amazon Aurora的架構有三大優勢："}]},{"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":"第一、Amazon Aurora天生就是計算和存儲分離的，它的底層數據庫存儲其實就是一個分佈式的存儲服務，我們可以認爲系統有一個巨大的共享存儲池在下面。"}]},{"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":"第二、Amazon Aurora的設計哲學總結起來就是Log is Database，所以數據庫的計算實例和底層的存儲之間得寫log，這樣的目的是減少計算實例和存儲之間的網絡帶寬需求，極大提升數據庫的性能。"}]},{"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":"第三、Amazon Aurora會把數據庫一部分的核心功能，像故障恢復、備份還原到分佈式的存儲層去解決，這些任務可以在後臺執行，不會影響前臺用戶的任務。"}]},{"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":"成本的節省在亞馬遜自身得以體現，作爲Oracle全球排名前幾的大客戶，亞馬遜電商應該能夠從甲骨文拿到非常好的折扣，但就算在折扣的基礎上，實際上從Oracle數據庫切換到亞馬遜雲科技的雲數據庫後，亞馬遜電商也節省了60%的成本，同時面向消費者端的應用程序的延遲降低了40%，數據庫管理支出減少了70%。"}]}]}