{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"大家好，我是悟空。","attrs":{}}]},{"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","text":"上一篇講到了 MySQL 和 NoSQL 的區別和優缺點","attrs":{}},{"type":"link","attrs":{"href":"https://mp.weixin.qq.com/s/6EqD8KBs3C9TZfXCfbOJ0w","title":"","type":null},"content":[{"type":"text","text":"《有了 MySQL，爲什麼還要 NoSQL》","attrs":{}}]},{"type":"text","text":"，這次我們來聊下分佈式場景下","attrs":{}},{"type":"codeinline","content":[{"type":"text","text":"數據庫","attrs":{}}],"attrs":{}},{"type":"text","text":"。首先我們還是來看下關係型和非關係型的數據庫的區別和特點。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"一、關係型 vs 非關係型","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"1.1 關係型","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"1.1.1 什麼是關係型？","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"關係型數據庫指的是使用","attrs":{}},{"type":"codeinline","content":[{"type":"text","text":"關係模型","attrs":{}}],"attrs":{}},{"type":"text","text":"（二維表格模型）來組織數據的數據庫，由二維表及其之間的聯繫所組成的一個數據組織。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"1.1.2 常見關係型數據庫","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"常見關係型數據庫管理系統（ORDBMS）：Oracle、MySql、Microsoft SQL Server、SQLite、PostgreSQ、IBM DB2。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"1.1.3 關係型的優勢","attrs":{}}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"採用二維表結構非常貼近正常開發邏輯。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"支持通用的SQL（結構化查詢語言）語句。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"豐富的完整性大大減少了數據冗餘和數據不一致的問題。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"可以用SQL句子多個表之間做非常繁雜的查詢；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"關係型數據庫提供對事務的支持。","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"1.1.4 關係型的不足之處","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"（1）存儲的是行記錄","attrs":{}},{"type":"text","text":"。","attrs":{}}]},{"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","text":"不能存儲數組、嵌套字段等格式的數據。","attrs":{}}]},{"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":"strong","attrs":{}}],"text":"（2）擴展表結構不方便","attrs":{}},{"type":"text","text":"。","attrs":{}}]},{"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","text":"操作不存在的列會報錯，而增加列又需要執行 SQL 語句纔行。而且修改時需要特別注意，因爲更新表時會長時間","attrs":{}},{"type":"codeinline","content":[{"type":"text","text":"鎖表","attrs":{}}],"attrs":{}},{"type":"text","text":"，這對線上環境可能造成嚴重影響。","attrs":{}}]},{"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":"strong","attrs":{}}],"text":"（3）佔用內存高","attrs":{}},{"type":"text","text":"。","attrs":{}}]},{"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","text":"關係型數據庫在對大量數據的表進行統計之類的運算時，佔用內存會很高，因爲它即使只針對某一列進行運算，也會將整行數據從存儲設備讀入內存。","attrs":{}}]},{"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":"strong","attrs":{}}],"text":"（4）全文搜索性能差","attrs":{}}]},{"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","text":"類似於 MySQL 的關係型數據庫，只能用 like 進行整表掃描的匹配，效率很低。現如今，有很多場景需要支持模糊匹配，而且必須支持高效查找。比如查詢包含關鍵字的日誌信息，又或者是根據某個商品關鍵字查詢商品列表。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"1.2 非關係型","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"1.2.1 什麼是非關係型？","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/b1/b17604aba45885448555af99f74e394e.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"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","text":"NoSQL(NoSQL = Not Only SQL )，意即\"不僅僅是SQL\"。","attrs":{}}]},{"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","text":"非關係型數據庫嚴格上不是一種數據庫，應該是一種數據結構化存儲方法的集合，可以是文檔或者鍵值對等。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"1.2.2 常見非關係型數據庫","attrs":{}}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"鍵值數據庫：Redis、Memcached、Riak。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"列族數據庫：Bigtable、HBase、Cassandra。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"文檔數據庫：MongoDB、CouchDB、MarkLogic。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"圖形數據庫：Neo4j、InfoGrid。","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"1.2.3 非關係型的優勢","attrs":{}}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"格式靈活：存儲數據的格式可以是key,value形式、文檔形式、圖片形式等等，文檔形式、圖片形式等等，使用靈活，應用場景廣泛，而關係型數據庫則只支持基礎類型。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"速度快：NoSQL 可以使用硬盤或者內存來存儲，而關係型數據庫只能使用硬盤；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"高擴展性；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"成本低：nosql數據庫部署簡單，基本都是開源軟件。","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"1.2.4 非關係型的不足之處","attrs":{}}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"不提供sql支持，學習和使用成本較高；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"無事務處理。MongoDB 4.0 已支持事務。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"數據結構相對複雜，複雜查詢方面稍欠。","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"二、分佈式數據庫","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"2.1 分佈式數據庫的定義","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"codeinline","content":[{"type":"text","text":"分佈式數據庫","attrs":{}}],"attrs":{}},{"type":"text","text":"其實沒有一個官方的定義，只是我們技術人員提出的一個約定俗成的說法。","attrs":{}}]},{"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","text":"在數據庫領域，當產品不斷演進逐漸被大家認識和認可後，就會成了一個標準，比如說微軟的 SQL Server 數據庫，其他數據庫都喜歡拿它作爲對比，那 SQL Server 數據庫就會成爲一個標準。","attrs":{}}]},{"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","text":"但是分佈式數據庫也是最近幾年才被大家提出，還是比較新的，也沒有參照。不過我們可以通過這些大廠大牛們總結的經驗來認識分佈式數據庫。","attrs":{}}]},{"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","text":"定義：分佈式數據庫就是用分佈式架構實現的數據庫。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"2.2 分佈式數據庫的優勢","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"分佈式一直是我研究的一個話題，現在很多流行的技術都用上了分佈式架構，比如微服務、消息隊列。","attrs":{}}]},{"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","text":"那爲什麼我們要用分佈式架構呢？簡單來說，就是用多機（機器）來橫向擴展單機的","attrs":{}},{"type":"codeinline","content":[{"type":"text","text":"性能","attrs":{}}],"attrs":{}},{"type":"text","text":"，另外一個很重要的原因就是分佈式的","attrs":{}},{"type":"codeinline","content":[{"type":"text","text":"可靠性","attrs":{}}],"attrs":{}},{"type":"text","text":"，比如多機備份、容災等。","attrs":{}}]},{"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","text":"那數據庫是不是也需要提升性能和保證可靠性呢？答案是肯定的。","attrs":{}}]},{"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","text":"哪些大廠在用分佈式數據庫？","attrs":{}}]},{"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","text":"每年雙 11，阿里就喜歡 show 一波交易戰績，其分佈式數據庫 OceanBase 功不可沒。頭部大廠如騰訊、字節跳動、美團也開始使用分佈式數據庫，還有各大銀行也上線了分佈式數據庫。","attrs":{}}]},{"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","text":"所以說分佈式數據庫是一種趨勢，如果業務場景要求高性能和高可靠，就可以考慮使用分佈式架構下的數據庫了。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"2.3 分佈式數據庫的特點","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"首先我們來看下數據庫按照交易類型區分的兩大場景：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"聯機交易（OLTP）","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"OLTP 是面向交易的處理過程，單筆交易的數據量小，但是要在很短的時間內給出結果，典型場景包括購物、繳費、轉賬等；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"聯機分析（OLAP）","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"OLAP 場景通常是基於大數據集的運算，典型場景包括生成個人年度賬單和企業財務報表等。","attrs":{}}]}]}],"attrs":{}},{"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","text":"OLTP 的特點是寫多讀少、低延時、高併發，那麼數據庫+分佈式在 OLTP 場景下會具有哪些特點呢？","attrs":{}}]},{"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":"strong","attrs":{}}],"text":"特點","attrs":{}},{"type":"text","text":"：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在寫多讀少的場景很強大。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"低延時的響應。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"支持高併發。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"支持海量存儲。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"高可靠性。","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"三、10 種分佈式數據庫","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.1 PingCAP 的 TiDB","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"開源 + 良好的社區運營，擁有超高人氣。","attrs":{}}]},{"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":"strong","attrs":{}}],"text":"定義","attrs":{}},{"type":"text","text":"：是一款同時支持在線事務處理與在線分析處理 (Hybrid Transactional and Analytical Processing, HTAP）的融合型分佈式數據庫產品，具備水平擴容或者縮容、金融級高可用、實時 HTAP、雲原生的分佈式數據庫、兼容 MySQL 5.7 協議和 MySQL 生態等重要特性。目標是爲用戶提供一站式 OLTP (Online Transactional Processing)、OLAP (Online Analytical Processing)、HTAP 解決方案。TiDB 適合高可用、強一致要求較高、數據規模較大等各種應用場景。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/81/81ac0f67869f305151a145acaa720d47.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"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","text":"TIDB 採用分層架構，有三種角色：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"TIDB：作爲 SQL 引擎。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"TiKV：作爲底層分佈式鍵值存儲。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"PD：承擔元數據管理和全局時鐘的職責。","attrs":{}}]}]}],"attrs":{}},{"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","text":"TiDB 的衍生項目：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Ti-Binlog、Ti-CDC 支持數據導出。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Ti-Operator 更方便地實現容器雲部署。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Chaos Mesh 支持混沌工程。","attrs":{}}]}]}],"attrs":{}},{"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","text":"缺點：不支持全球化部署，這爲跨地域大規模集羣應用 TiDB 設置了障礙。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.2 Google 的 Spanner","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Spanner是谷歌公司研發的、可擴展的、多版本、全球分佈式、同步複製數據庫。它支持外部一致性的分佈式事務。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/7c/7ccdf211f0fc32a158fa983c36af713f.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"codeinline","content":[{"type":"text","text":"F1","attrs":{}}],"attrs":{}},{"type":"text","text":" 主要作爲 SQL 引擎","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"codeinline","content":[{"type":"text","text":"Spanner","attrs":{}}],"attrs":{}},{"type":"text","text":" 主要負責事務一致性、複製機制、可擴展存儲等。","attrs":{}}]},{"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","text":"Spanner 架構中的核心處理模塊是 Spanserver，","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/19/198bbf1dcd7be081d3c3787ef6f09e7d.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"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","text":"Spanserver 的核心工作有三部分：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"基於 ","attrs":{}},{"type":"codeinline","content":[{"type":"text","text":"Paxos","attrs":{}}],"attrs":{}},{"type":"text","text":" 協議的數據複製。Paxos 協議可以看我之前寫的一篇文章：","attrs":{}},{"type":"link","attrs":{"href":"https://mp.weixin.qq.com/s/Cmb7oEX4mot4HwkyPLJMqQ","title":"","type":null},"content":[{"type":"text","text":"《用三國殺講分佈式算法，舒適了吧》","attrs":{}}]}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"基於 ","attrs":{}},{"type":"codeinline","content":[{"type":"text","text":"Tablet","attrs":{}}],"attrs":{}},{"type":"text","text":" 的分片管理。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"基於 ","attrs":{}},{"type":"codeinline","content":[{"type":"text","text":"2PC","attrs":{}}],"attrs":{}},{"type":"text","text":" 的事務一致性管理。2PC 協議可以看我之前寫的一篇文章：","attrs":{}},{"type":"link","attrs":{"href":"https://mp.weixin.qq.com/s/6g53y7Mv2sqRgFw9DF5Iww","title":"","type":null},"content":[{"type":"text","text":"《用太極拳講分佈式理論，真舒服》","attrs":{}}]}]}]}],"attrs":{}},{"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","text":"2017 年，F1 和 Spanner 被拆分了，不在是綁定關係。原理如如下：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/52/526ddab4467b1fb7a05be01369f98010.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.3 CockroachDB 蟑螂數據庫","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"CockroachDB (蟑螂數據庫）是一個可伸縮的、支持地理位置處理、支持事務處理的數據存儲系統。","attrs":{}}]},{"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":"strong","attrs":{}}],"text":"爲什麼叫做蟑螂？","attrs":{}}]},{"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","text":"因爲這個數據庫只要損壞的節點不超過總數一半，那麼集羣仍然可以正常工作，生命力超強。","attrs":{}}]},{"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","text":"通過分佈式一致性算法實例來調節確保一致性，它所選擇使用Raft一致性算法。所有的一致性狀態存在於RocksDB中。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/df/df2f9719cbddaa3e27d2df182479f0fa.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"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","text":"Cockroach 是一個分佈式的 SQL 數據庫。首要設計目標就是 ","attrs":{}},{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"可擴展性","attrs":{}},{"type":"text","text":"，","attrs":{}},{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"強一致性","attrs":{}},{"type":"text","text":"，","attrs":{}},{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"可存活性","attrs":{}},{"type":"text","text":"，就像它的名字一樣。Cockroach 的目標是在無人工干預的情況下，以極小的中斷時間容忍磁盤，主機，機架甚至 ","attrs":{}},{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"數據中心災難","attrs":{}},{"type":"text","text":" 。Cockroach 的節點是對等的，其中一個設計目標是以最少配置加無依賴，部署去中心化的對等節點。中文社區地址：cockroachdb-cn。","attrs":{}}]},{"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","text":"CockroachDB 提供兩種不同的的事務特性，包括快照隔離（snapshot isolation，簡稱SI）和順序的快照隔離（SSI）語義，後者是默認的隔離級別。","attrs":{}}]},{"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","text":"CockroachDB 是一個分佈式的K/V數據倉庫，支持ACID事務，多版本值存儲是其首要特性。主要的設計目標是全球一致性和可靠性，從蟑螂的命名上是就能看出這點。蟑螂數據庫能處理磁盤、物理機器、機架甚至數據中心失效情況下最小延遲的服務中斷；整個失效過程無需人工干預。蟑螂的節點是均衡的，其設計目標是同質部署（只有一個二進制包）且最小配置。","attrs":{}}]},{"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","text":"CockroachDB 和 TiDB、YugabyteDB 都公開聲稱設計靈感來自 Spanner，所以往往會被認爲是同構的產品。CockroachDB 和 TiDB，經常會被大家拿來比較。","attrs":{}}]},{"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","text":"區別：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"CockroachDB 採用了標準的 P2P 架構，只要損壞的節點不超過總數一半，那麼集羣仍然可以正常工作。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"CockroachDB 支持全球化部署，因爲它採用了混合邏輯時鐘（HLC），所以能夠在全球物理範圍下做到數據一致性。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"分片管理機制的不同。","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.4 YugabyteDB","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在架構上和 CockroachDB 有很多相似之處，比如支持全球化部署，採用混合邏輯時鐘（HLC），基於 Percolator 的事務模型，兼容 PostgreSQL 協議。","attrs":{}}]},{"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","text":"由於高度的相似性，YugabyteDB 與 CockroachDB 的競爭表現得非常激烈。","attrs":{}}]},{"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","text":"Yugabyte 採用兩層架構：查詢層和存儲層。不過這個架構僅僅是邏輯上的，部署結構中，這兩層都位於 TServer 進程中。這一點和 TiDB 不同。","attrs":{}}]},{"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","text":"Yugabyte 的查詢層支持同時 SQL 和 CQL 兩種 API，其中 CQL 是兼容 Cassandra 的一種方言語法，對應於文檔數據庫的存儲模型；而 SQL API 是直接基於 PostgresQL 魔改的，能比較好地兼容 PG 語法，","attrs":{}}]},{"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","text":"Yugabyte 的存儲層纔是重頭戲。其中 TServer 負責存儲 tablet，每個 tablet 對應一個 Raft Group，分佈在三個不同的節點上，以此保證高可用性。Master 負責元數據管理，除了 tablet 的位置信息，還包括表結構等信息。Master 本身也依靠 Raft 實現高可用。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/39/3967104021069f9527d2e20c09741f0c.jpeg","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.5 阿里OceanBase","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"OceanBase是由螞蟻集團完全自主研發的金融級分佈式關係數據庫，始創於2010年。OceanBase具有數據強一致、高可用、高性能、在線擴展、高度兼容SQL標準和主流關係數據庫、低成本等特點。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.6 騰訊的 TDSQL","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"TDSQL 的節點都是用的 MySQL。它是採用分佈式集羣架構（如下圖），這種集羣架構具有較高的靈活性，簡化了各個 節點之間的通信機制，也簡化了對於硬件的需求。這不僅意味着 TDSQL 的關係型實例、分 布式實例、分析性實例可以混合部署在同一集羣中，也意味着即使是簡單的 x86 服務器，也 可以搭建出類似於小型機、共享存儲等一樣穩定可靠的數據庫。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/1a/1adb44a5a89862fed37c2b87bf4198bf.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.7 中興通訊的 GoldenDB","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"GoldenDB 幾乎是國內銀行業應用規模最大的分佈式數據庫，和 TDSQL 同樣在數據節點上選擇了 MySQL，但全局時鐘節點的增加使它稱爲一個標準的 PGXC 架構。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/17/17aa0420fd28ae2fbc7b5a3f822f8cd3.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.8 騰訊的 TBase","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"TBase","attrs":{}},{"type":"text","text":" 是騰訊數據平臺團隊在開源的 PostgreSQL 基礎上研發的企業級分佈式 HTAP 數據庫管理系統：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"具備高性能可擴展的分佈式事務能力，支持 RC 和 RR 兩種隔離級別；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"通過安全、管理、審計三權分立體系，提供全方位的數據安全保證機制；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"支持高性能分區表，可使得數據檢索效率成倍提升；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"SQL 方面兼容 2003 標準、PostgreSQL 語法和常用 Oracle 函數&數據類型、窗口函數等；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"提供大小商戶數據分離、冷熱數據分離等高效的數據治理能力","attrs":{}}]}]}],"attrs":{}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/b7/b7825f6661c852a648f4766303db9cae.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"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","text":"集羣中有三種節點類型，各自承擔不同的功能，通過網絡連接成爲一個系統。這三種節點類型分別是：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"**Coordinator：**協調節點，對外提供接口，負責數據的分發和查詢規劃，多個節點位置對等，每個節點都提供相同的數據庫視圖，CN 存儲系統的全局元數據。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"**Datanode：**處理存儲本節點相關的元數據，每個節點還存儲數據的一個分片。在功能上，DN 節點負責完成執行協調節點分發的執行請求。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"GTM:","attrs":{}},{"type":"text","text":" 全局事務管理器(Global transaction manager.)，負責管理集羣事務信息，同時管理集羣的全局對象，比如序列，除此之外 GTM 上不提供其他的功能。","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.9 VoltDB","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"VoltDB 官網提供的簡介： VoltDB是全球最快的內存型數據庫，它繼承了傳統關係數據庫的強一致性要求，又提供了互聯網雲上部署的能力和分佈式 數據庫的橫向擴展能力。VoltDB通過將數據庫全部保存在內存中的方法，消除了大量的數據和日誌的磁盤存取操作，通過單線程的方式，消除了磁盤鎖和記錄鎖；通過數據庫分片技術，讓數據庫支持高併發請求；通過分佈式集羣支持數據庫橫向擴展。其查詢速度達到傳統數據庫的100倍以上。","attrs":{}}]},{"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","text":"2019 年正式閉源，變爲純商業化產品。而同時，VoltDB 在國內也沒有建立完備的服務支持體系，這在很大程度上影響到它的推廣。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.10 巨杉 SequoiaDB","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"SequoiaDB 巨杉數據庫是一款開源的金融級分佈式關係型數據庫，主要面對高併發聯機交易型場景提供高性能、可靠穩定以及無限水平擴展的數據庫服務。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/ab/abfe54441de6a60bc3321282356e987e.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"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","text":"用戶可以在 SequoiaDB 巨杉數據庫中創建多種類型的數據庫實例，以滿足上層不同應用程序各自的需求。","attrs":{}}]},{"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","text":"SequoiaDB 巨杉數據庫支持 MySQL、PostgreSQL、SparkSQL 和 MariaDB 四種關係型數據庫實例、類 MongoDB 的 JSON 文檔類數據庫實例、以及 S3 對象存儲與 POSIX 文件系統的非結構化數據實例。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/4a/4a3a49da74326d0ede5e3236d52481de.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"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","text":"SequoiaDB 巨杉數據庫存儲引擎採用分佈式架構。集羣中的每個節點爲一個獨立進程，節點之間採用 TCP/IP 協議進行通訊。","attrs":{}}]},{"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","text":"同一個操作系統可以部署多個節點，節點之間採用不同的端口進行區分。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/23/23e8d00f760ac961eab65568279fecc6.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"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","text":"好了，對於分佈式數據庫，如果你也有分佈式數據庫的使用經驗，歡迎留言～","attrs":{}}]},{"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":"size","attrs":{"size":9}}],"text":"https://www.zhihu.com/question/24225007/answer/1707736658","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"https://www.zhihu.com/question/24225007/answer/1326259742","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"https://time.geekbang.org/column/article/296558","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"https://docs.pingcap.com/zh/tidb/stable","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"https://storage.googleapis.com/pub-tools-public-publication-data/pdf/41344.pdf","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"http://storage.googleapis.com/pub-tools-public-publication-data/pdf/39966.pdf","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"http://vldb.org/pvldb/vol11/p1835-samwel.pdf","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"http://ericfu.me/yugabyte-db-introduction/","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"https://blog.csdn.net/duan_zhihua/article/details/88549173","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"https://www.voltdb.com/","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"http://doc.sequoiadb.com/cn/sequoiadb-cat_id-1558957225-edition_id-500","attrs":{}}]},{"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","text":"歡迎關注我的公衆號：「","attrs":{}},{"type":"text","marks":[{"type":"color","attrs":{"color":"#FF7021","name":"user"}}],"text":"悟空聊架構","attrs":{}},{"type":"text","text":"」","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"作者簡介：","attrs":{}},{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"8 年互聯網職場老兵｜全棧工程師｜90 後超級奶爸｜開源踐行者｜公衆號萬粉原創號主","attrs":{}},{"type":"text","text":"。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"藍橋簽約作者，著有《JVM 性能調優實戰》專欄，手寫了一套 7 萬字 SpringCloud 實戰總結和 3 萬字分佈式算法總結。歡迎關注我的公衆號「悟空聊架構」，免費獲取資料學習。","attrs":{}}]}],"attrs":{}}]}
帶你遨遊銀河系的十種分佈式數據庫
{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"大家好，我是悟空。","attrs":{}}]},{"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","text":"上一篇講到了 MySQL 和 NoSQL 的區別和優缺點","attrs":{}},{"type":"link","attrs":{"href":"https://mp.weixin.qq.com/s/6EqD8KBs3C9TZfXCfbOJ0w","title":"","type":null},"content":[{"type":"text","text":"《有了 MySQL，爲什麼還要 NoSQL》","attrs":{}}]},{"type":"text","text":"，這次我們來聊下分佈式場景下","attrs":{}},{"type":"codeinline","content":[{"type":"text","text":"數據庫","attrs":{}}],"attrs":{}},{"type":"text","text":"。首先我們還是來看下關係型和非關係型的數據庫的區別和特點。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"一、關係型 vs 非關係型","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"1.1 關係型","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"1.1.1 什麼是關係型？","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"關係型數據庫指的是使用","attrs":{}},{"type":"codeinline","content":[{"type":"text","text":"關係模型","attrs":{}}],"attrs":{}},{"type":"text","text":"（二維表格模型）來組織數據的數據庫，由二維表及其之間的聯繫所組成的一個數據組織。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"1.1.2 常見關係型數據庫","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"常見關係型數據庫管理系統（ORDBMS）：Oracle、MySql、Microsoft SQL Server、SQLite、PostgreSQ、IBM DB2。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"1.1.3 關係型的優勢","attrs":{}}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"採用二維表結構非常貼近正常開發邏輯。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"支持通用的SQL（結構化查詢語言）語句。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"豐富的完整性大大減少了數據冗餘和數據不一致的問題。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"可以用SQL句子多個表之間做非常繁雜的查詢；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"關係型數據庫提供對事務的支持。","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"1.1.4 關係型的不足之處","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"（1）存儲的是行記錄","attrs":{}},{"type":"text","text":"。","attrs":{}}]},{"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","text":"不能存儲數組、嵌套字段等格式的數據。","attrs":{}}]},{"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":"strong","attrs":{}}],"text":"（2）擴展表結構不方便","attrs":{}},{"type":"text","text":"。","attrs":{}}]},{"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","text":"操作不存在的列會報錯，而增加列又需要執行 SQL 語句纔行。而且修改時需要特別注意，因爲更新表時會長時間","attrs":{}},{"type":"codeinline","content":[{"type":"text","text":"鎖表","attrs":{}}],"attrs":{}},{"type":"text","text":"，這對線上環境可能造成嚴重影響。","attrs":{}}]},{"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":"strong","attrs":{}}],"text":"（3）佔用內存高","attrs":{}},{"type":"text","text":"。","attrs":{}}]},{"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","text":"關係型數據庫在對大量數據的表進行統計之類的運算時，佔用內存會很高，因爲它即使只針對某一列進行運算，也會將整行數據從存儲設備讀入內存。","attrs":{}}]},{"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":"strong","attrs":{}}],"text":"（4）全文搜索性能差","attrs":{}}]},{"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","text":"類似於 MySQL 的關係型數據庫，只能用 like 進行整表掃描的匹配，效率很低。現如今，有很多場景需要支持模糊匹配，而且必須支持高效查找。比如查詢包含關鍵字的日誌信息，又或者是根據某個商品關鍵字查詢商品列表。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"1.2 非關係型","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"1.2.1 什麼是非關係型？","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/b1/b17604aba45885448555af99f74e394e.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"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","text":"NoSQL(NoSQL = Not Only SQL )，意即\"不僅僅是SQL\"。","attrs":{}}]},{"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","text":"非關係型數據庫嚴格上不是一種數據庫，應該是一種數據結構化存儲方法的集合，可以是文檔或者鍵值對等。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"1.2.2 常見非關係型數據庫","attrs":{}}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"鍵值數據庫：Redis、Memcached、Riak。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"列族數據庫：Bigtable、HBase、Cassandra。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"文檔數據庫：MongoDB、CouchDB、MarkLogic。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"圖形數據庫：Neo4j、InfoGrid。","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"1.2.3 非關係型的優勢","attrs":{}}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"格式靈活：存儲數據的格式可以是key,value形式、文檔形式、圖片形式等等，文檔形式、圖片形式等等，使用靈活，應用場景廣泛，而關係型數據庫則只支持基礎類型。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"速度快：NoSQL 可以使用硬盤或者內存來存儲，而關係型數據庫只能使用硬盤；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"高擴展性；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"成本低：nosql數據庫部署簡單，基本都是開源軟件。","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"1.2.4 非關係型的不足之處","attrs":{}}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"不提供sql支持，學習和使用成本較高；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"無事務處理。MongoDB 4.0 已支持事務。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"數據結構相對複雜，複雜查詢方面稍欠。","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"二、分佈式數據庫","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"2.1 分佈式數據庫的定義","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"codeinline","content":[{"type":"text","text":"分佈式數據庫","attrs":{}}],"attrs":{}},{"type":"text","text":"其實沒有一個官方的定義，只是我們技術人員提出的一個約定俗成的說法。","attrs":{}}]},{"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","text":"在數據庫領域，當產品不斷演進逐漸被大家認識和認可後，就會成了一個標準，比如說微軟的 SQL Server 數據庫，其他數據庫都喜歡拿它作爲對比，那 SQL Server 數據庫就會成爲一個標準。","attrs":{}}]},{"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","text":"但是分佈式數據庫也是最近幾年才被大家提出，還是比較新的，也沒有參照。不過我們可以通過這些大廠大牛們總結的經驗來認識分佈式數據庫。","attrs":{}}]},{"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","text":"定義：分佈式數據庫就是用分佈式架構實現的數據庫。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"2.2 分佈式數據庫的優勢","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"分佈式一直是我研究的一個話題，現在很多流行的技術都用上了分佈式架構，比如微服務、消息隊列。","attrs":{}}]},{"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","text":"那爲什麼我們要用分佈式架構呢？簡單來說，就是用多機（機器）來橫向擴展單機的","attrs":{}},{"type":"codeinline","content":[{"type":"text","text":"性能","attrs":{}}],"attrs":{}},{"type":"text","text":"，另外一個很重要的原因就是分佈式的","attrs":{}},{"type":"codeinline","content":[{"type":"text","text":"可靠性","attrs":{}}],"attrs":{}},{"type":"text","text":"，比如多機備份、容災等。","attrs":{}}]},{"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","text":"那數據庫是不是也需要提升性能和保證可靠性呢？答案是肯定的。","attrs":{}}]},{"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","text":"哪些大廠在用分佈式數據庫？","attrs":{}}]},{"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","text":"每年雙 11，阿里就喜歡 show 一波交易戰績，其分佈式數據庫 OceanBase 功不可沒。頭部大廠如騰訊、字節跳動、美團也開始使用分佈式數據庫，還有各大銀行也上線了分佈式數據庫。","attrs":{}}]},{"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","text":"所以說分佈式數據庫是一種趨勢，如果業務場景要求高性能和高可靠，就可以考慮使用分佈式架構下的數據庫了。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"2.3 分佈式數據庫的特點","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"首先我們來看下數據庫按照交易類型區分的兩大場景：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"聯機交易（OLTP）","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"OLTP 是面向交易的處理過程，單筆交易的數據量小，但是要在很短的時間內給出結果，典型場景包括購物、繳費、轉賬等；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"聯機分析（OLAP）","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"OLAP 場景通常是基於大數據集的運算，典型場景包括生成個人年度賬單和企業財務報表等。","attrs":{}}]}]}],"attrs":{}},{"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","text":"OLTP 的特點是寫多讀少、低延時、高併發，那麼數據庫+分佈式在 OLTP 場景下會具有哪些特點呢？","attrs":{}}]},{"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":"strong","attrs":{}}],"text":"特點","attrs":{}},{"type":"text","text":"：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在寫多讀少的場景很強大。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"低延時的響應。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"支持高併發。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"支持海量存儲。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"高可靠性。","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"三、10 種分佈式數據庫","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.1 PingCAP 的 TiDB","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"開源 + 良好的社區運營，擁有超高人氣。","attrs":{}}]},{"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":"strong","attrs":{}}],"text":"定義","attrs":{}},{"type":"text","text":"：是一款同時支持在線事務處理與在線分析處理 (Hybrid Transactional and Analytical Processing, HTAP）的融合型分佈式數據庫產品，具備水平擴容或者縮容、金融級高可用、實時 HTAP、雲原生的分佈式數據庫、兼容 MySQL 5.7 協議和 MySQL 生態等重要特性。目標是爲用戶提供一站式 OLTP (Online Transactional Processing)、OLAP (Online Analytical Processing)、HTAP 解決方案。TiDB 適合高可用、強一致要求較高、數據規模較大等各種應用場景。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/81/81ac0f67869f305151a145acaa720d47.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"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","text":"TIDB 採用分層架構，有三種角色：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"TIDB：作爲 SQL 引擎。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"TiKV：作爲底層分佈式鍵值存儲。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"PD：承擔元數據管理和全局時鐘的職責。","attrs":{}}]}]}],"attrs":{}},{"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","text":"TiDB 的衍生項目：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Ti-Binlog、Ti-CDC 支持數據導出。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Ti-Operator 更方便地實現容器雲部署。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Chaos Mesh 支持混沌工程。","attrs":{}}]}]}],"attrs":{}},{"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","text":"缺點：不支持全球化部署，這爲跨地域大規模集羣應用 TiDB 設置了障礙。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.2 Google 的 Spanner","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Spanner是谷歌公司研發的、可擴展的、多版本、全球分佈式、同步複製數據庫。它支持外部一致性的分佈式事務。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/7c/7ccdf211f0fc32a158fa983c36af713f.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"codeinline","content":[{"type":"text","text":"F1","attrs":{}}],"attrs":{}},{"type":"text","text":" 主要作爲 SQL 引擎","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"codeinline","content":[{"type":"text","text":"Spanner","attrs":{}}],"attrs":{}},{"type":"text","text":" 主要負責事務一致性、複製機制、可擴展存儲等。","attrs":{}}]},{"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","text":"Spanner 架構中的核心處理模塊是 Spanserver，","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/19/198bbf1dcd7be081d3c3787ef6f09e7d.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"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","text":"Spanserver 的核心工作有三部分：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"基於 ","attrs":{}},{"type":"codeinline","content":[{"type":"text","text":"Paxos","attrs":{}}],"attrs":{}},{"type":"text","text":" 協議的數據複製。Paxos 協議可以看我之前寫的一篇文章：","attrs":{}},{"type":"link","attrs":{"href":"https://mp.weixin.qq.com/s/Cmb7oEX4mot4HwkyPLJMqQ","title":"","type":null},"content":[{"type":"text","text":"《用三國殺講分佈式算法，舒適了吧》","attrs":{}}]}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"基於 ","attrs":{}},{"type":"codeinline","content":[{"type":"text","text":"Tablet","attrs":{}}],"attrs":{}},{"type":"text","text":" 的分片管理。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"基於 ","attrs":{}},{"type":"codeinline","content":[{"type":"text","text":"2PC","attrs":{}}],"attrs":{}},{"type":"text","text":" 的事務一致性管理。2PC 協議可以看我之前寫的一篇文章：","attrs":{}},{"type":"link","attrs":{"href":"https://mp.weixin.qq.com/s/6g53y7Mv2sqRgFw9DF5Iww","title":"","type":null},"content":[{"type":"text","text":"《用太極拳講分佈式理論，真舒服》","attrs":{}}]}]}]}],"attrs":{}},{"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","text":"2017 年，F1 和 Spanner 被拆分了，不在是綁定關係。原理如如下：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/52/526ddab4467b1fb7a05be01369f98010.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.3 CockroachDB 蟑螂數據庫","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"CockroachDB (蟑螂數據庫）是一個可伸縮的、支持地理位置處理、支持事務處理的數據存儲系統。","attrs":{}}]},{"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":"strong","attrs":{}}],"text":"爲什麼叫做蟑螂？","attrs":{}}]},{"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","text":"因爲這個數據庫只要損壞的節點不超過總數一半，那麼集羣仍然可以正常工作，生命力超強。","attrs":{}}]},{"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","text":"通過分佈式一致性算法實例來調節確保一致性，它所選擇使用Raft一致性算法。所有的一致性狀態存在於RocksDB中。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/df/df2f9719cbddaa3e27d2df182479f0fa.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"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","text":"Cockroach 是一個分佈式的 SQL 數據庫。首要設計目標就是 ","attrs":{}},{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"可擴展性","attrs":{}},{"type":"text","text":"，","attrs":{}},{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"強一致性","attrs":{}},{"type":"text","text":"，","attrs":{}},{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"可存活性","attrs":{}},{"type":"text","text":"，就像它的名字一樣。Cockroach 的目標是在無人工干預的情況下，以極小的中斷時間容忍磁盤，主機，機架甚至 ","attrs":{}},{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"數據中心災難","attrs":{}},{"type":"text","text":" 。Cockroach 的節點是對等的，其中一個設計目標是以最少配置加無依賴，部署去中心化的對等節點。中文社區地址：cockroachdb-cn。","attrs":{}}]},{"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","text":"CockroachDB 提供兩種不同的的事務特性，包括快照隔離（snapshot isolation，簡稱SI）和順序的快照隔離（SSI）語義，後者是默認的隔離級別。","attrs":{}}]},{"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","text":"CockroachDB 是一個分佈式的K/V數據倉庫，支持ACID事務，多版本值存儲是其首要特性。主要的設計目標是全球一致性和可靠性，從蟑螂的命名上是就能看出這點。蟑螂數據庫能處理磁盤、物理機器、機架甚至數據中心失效情況下最小延遲的服務中斷；整個失效過程無需人工干預。蟑螂的節點是均衡的，其設計目標是同質部署（只有一個二進制包）且最小配置。","attrs":{}}]},{"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","text":"CockroachDB 和 TiDB、YugabyteDB 都公開聲稱設計靈感來自 Spanner，所以往往會被認爲是同構的產品。CockroachDB 和 TiDB，經常會被大家拿來比較。","attrs":{}}]},{"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","text":"區別：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"CockroachDB 採用了標準的 P2P 架構，只要損壞的節點不超過總數一半，那麼集羣仍然可以正常工作。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"CockroachDB 支持全球化部署，因爲它採用了混合邏輯時鐘（HLC），所以能夠在全球物理範圍下做到數據一致性。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"分片管理機制的不同。","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.4 YugabyteDB","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在架構上和 CockroachDB 有很多相似之處，比如支持全球化部署，採用混合邏輯時鐘（HLC），基於 Percolator 的事務模型，兼容 PostgreSQL 協議。","attrs":{}}]},{"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","text":"由於高度的相似性，YugabyteDB 與 CockroachDB 的競爭表現得非常激烈。","attrs":{}}]},{"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","text":"Yugabyte 採用兩層架構：查詢層和存儲層。不過這個架構僅僅是邏輯上的，部署結構中，這兩層都位於 TServer 進程中。這一點和 TiDB 不同。","attrs":{}}]},{"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","text":"Yugabyte 的查詢層支持同時 SQL 和 CQL 兩種 API，其中 CQL 是兼容 Cassandra 的一種方言語法，對應於文檔數據庫的存儲模型；而 SQL API 是直接基於 PostgresQL 魔改的，能比較好地兼容 PG 語法，","attrs":{}}]},{"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","text":"Yugabyte 的存儲層纔是重頭戲。其中 TServer 負責存儲 tablet，每個 tablet 對應一個 Raft Group，分佈在三個不同的節點上，以此保證高可用性。Master 負責元數據管理，除了 tablet 的位置信息，還包括表結構等信息。Master 本身也依靠 Raft 實現高可用。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/39/3967104021069f9527d2e20c09741f0c.jpeg","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.5 阿里OceanBase","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"OceanBase是由螞蟻集團完全自主研發的金融級分佈式關係數據庫，始創於2010年。OceanBase具有數據強一致、高可用、高性能、在線擴展、高度兼容SQL標準和主流關係數據庫、低成本等特點。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.6 騰訊的 TDSQL","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"TDSQL 的節點都是用的 MySQL。它是採用分佈式集羣架構（如下圖），這種集羣架構具有較高的靈活性，簡化了各個 節點之間的通信機制，也簡化了對於硬件的需求。這不僅意味着 TDSQL 的關係型實例、分 布式實例、分析性實例可以混合部署在同一集羣中，也意味着即使是簡單的 x86 服務器，也 可以搭建出類似於小型機、共享存儲等一樣穩定可靠的數據庫。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/1a/1adb44a5a89862fed37c2b87bf4198bf.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.7 中興通訊的 GoldenDB","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"GoldenDB 幾乎是國內銀行業應用規模最大的分佈式數據庫，和 TDSQL 同樣在數據節點上選擇了 MySQL，但全局時鐘節點的增加使它稱爲一個標準的 PGXC 架構。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/17/17aa0420fd28ae2fbc7b5a3f822f8cd3.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.8 騰訊的 TBase","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"TBase","attrs":{}},{"type":"text","text":" 是騰訊數據平臺團隊在開源的 PostgreSQL 基礎上研發的企業級分佈式 HTAP 數據庫管理系統：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"具備高性能可擴展的分佈式事務能力，支持 RC 和 RR 兩種隔離級別；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"通過安全、管理、審計三權分立體系，提供全方位的數據安全保證機制；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"支持高性能分區表，可使得數據檢索效率成倍提升；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"SQL 方面兼容 2003 標準、PostgreSQL 語法和常用 Oracle 函數&數據類型、窗口函數等；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"提供大小商戶數據分離、冷熱數據分離等高效的數據治理能力","attrs":{}}]}]}],"attrs":{}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/b7/b7825f6661c852a648f4766303db9cae.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"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","text":"集羣中有三種節點類型，各自承擔不同的功能，通過網絡連接成爲一個系統。這三種節點類型分別是：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"**Coordinator：**協調節點，對外提供接口，負責數據的分發和查詢規劃，多個節點位置對等，每個節點都提供相同的數據庫視圖，CN 存儲系統的全局元數據。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"**Datanode：**處理存儲本節點相關的元數據，每個節點還存儲數據的一個分片。在功能上，DN 節點負責完成執行協調節點分發的執行請求。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"GTM:","attrs":{}},{"type":"text","text":" 全局事務管理器(Global transaction manager.)，負責管理集羣事務信息，同時管理集羣的全局對象，比如序列，除此之外 GTM 上不提供其他的功能。","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.9 VoltDB","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"VoltDB 官網提供的簡介： VoltDB是全球最快的內存型數據庫，它繼承了傳統關係數據庫的強一致性要求，又提供了互聯網雲上部署的能力和分佈式 數據庫的橫向擴展能力。VoltDB通過將數據庫全部保存在內存中的方法，消除了大量的數據和日誌的磁盤存取操作，通過單線程的方式，消除了磁盤鎖和記錄鎖；通過數據庫分片技術，讓數據庫支持高併發請求；通過分佈式集羣支持數據庫橫向擴展。其查詢速度達到傳統數據庫的100倍以上。","attrs":{}}]},{"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","text":"2019 年正式閉源，變爲純商業化產品。而同時，VoltDB 在國內也沒有建立完備的服務支持體系，這在很大程度上影響到它的推廣。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.10 巨杉 SequoiaDB","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"SequoiaDB 巨杉數據庫是一款開源的金融級分佈式關係型數據庫，主要面對高併發聯機交易型場景提供高性能、可靠穩定以及無限水平擴展的數據庫服務。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/ab/abfe54441de6a60bc3321282356e987e.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"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","text":"用戶可以在 SequoiaDB 巨杉數據庫中創建多種類型的數據庫實例，以滿足上層不同應用程序各自的需求。","attrs":{}}]},{"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","text":"SequoiaDB 巨杉數據庫支持 MySQL、PostgreSQL、SparkSQL 和 MariaDB 四種關係型數據庫實例、類 MongoDB 的 JSON 文檔類數據庫實例、以及 S3 對象存儲與 POSIX 文件系統的非結構化數據實例。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/4a/4a3a49da74326d0ede5e3236d52481de.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"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","text":"SequoiaDB 巨杉數據庫存儲引擎採用分佈式架構。集羣中的每個節點爲一個獨立進程，節點之間採用 TCP/IP 協議進行通訊。","attrs":{}}]},{"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","text":"同一個操作系統可以部署多個節點，節點之間採用不同的端口進行區分。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/23/23e8d00f760ac961eab65568279fecc6.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"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","text":"好了，對於分佈式數據庫，如果你也有分佈式數據庫的使用經驗，歡迎留言～","attrs":{}}]},{"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":"size","attrs":{"size":9}}],"text":"https://www.zhihu.com/question/24225007/answer/1707736658","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"https://www.zhihu.com/question/24225007/answer/1326259742","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"https://time.geekbang.org/column/article/296558","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"https://docs.pingcap.com/zh/tidb/stable","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"https://storage.googleapis.com/pub-tools-public-publication-data/pdf/41344.pdf","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"http://storage.googleapis.com/pub-tools-public-publication-data/pdf/39966.pdf","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"http://vldb.org/pvldb/vol11/p1835-samwel.pdf","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"http://ericfu.me/yugabyte-db-introduction/","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"https://blog.csdn.net/duan_zhihua/article/details/88549173","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"https://www.voltdb.com/","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":9}}],"text":"http://doc.sequoiadb.com/cn/sequoiadb-cat_id-1558957225-edition_id-500","attrs":{}}]},{"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","text":"歡迎關注我的公衆號：「","attrs":{}},{"type":"text","marks":[{"type":"color","attrs":{"color":"#FF7021","name":"user"}}],"text":"悟空聊架構","attrs":{}},{"type":"text","text":"」","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"作者簡介：","attrs":{}},{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"8 年互聯網職場老兵｜全棧工程師｜90 後超級奶爸｜開源踐行者｜公衆號萬粉原創號主","attrs":{}},{"type":"text","text":"。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"藍橋簽約作者，著有《JVM 性能調優實戰》專欄，手寫了一套 7 萬字 SpringCloud 實戰總結和 3 萬字分佈式算法總結。歡迎關注我的公衆號「悟空聊架構」，免費獲取資料學習。","attrs":{}}]}],"attrs":{}}]}
發表評論
所有評論
還沒有人評論，想成為第一個評論的人麼? 請在上方評論欄輸入並且點擊發布.
相關文章