{"type":"doc","content":[{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"每日一句","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","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#40A9FF","name":"blue"}},{"type":"strong","attrs":{}}],"text":"愛一個人要適度，和愛人的距離也要適度。愛他在不遠不近的距離，他便會對你不離不棄。","attrs":{}}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"數據庫瓶頸","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#000000","name":"black"}}],"text":" ","attrs":{}},{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#000000","name":"black"}},{"type":"strong","attrs":{}}],"text":"不管是IO瓶頸，還是CPU瓶頸，最終都會導致數據庫的活躍連接數增加，進而逼近甚至達到數據庫可承載活躍連接數的閾值。在業務Service來看就是，可用數據庫連接少甚至無連接可用。接下來就可以想象了吧（併發量、吞吐量、崩潰）。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"IO瓶頸","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#000000","name":"black"}},{"type":"strong","attrs":{}}],"text":" 第一種：磁盤讀IO瓶頸，熱點數據太多，數據庫緩存放不下，每次查詢時會產生大量的IO，降低查詢速度 -> 分庫和垂直分表","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#000000","name":"black"}},{"type":"strong","attrs":{}}],"text":" 第二種：網絡IO瓶頸，請求的數據太多，網絡帶寬不夠 ->分庫","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"CPU瓶頸","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" ","attrs":{}},{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":" ","attrs":{}},{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#000000","name":"black"}},{"type":"strong","attrs":{}}],"text":"第一種：SQL問題，如SQL中包含join，group by，order by，非索引字段條件查詢等，增加CPU運算的操作 -> SQL優化，建立合適的索引，在業務Service層進行業務計算。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#000000","name":"black"}},{"type":"strong","attrs":{}}],"text":" 第二種：單表數據量太大，查詢時掃描的行太多，SQL效率低，CPU率先出現瓶頸 -> 水平分表。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"分庫分表","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"水平分庫","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/7e/7e3611be103dc7bf850abf51a57dc40c.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":true,"pastePass":true}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"概念","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":" 以字段爲依據，按照一定策略（hash、range等），將一個庫中的數據拆分到多個庫中。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"結果","attrs":{}}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"每個庫的結構都一樣；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"每個庫的數據都不一樣，沒有交集；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"所有庫的並集是全量數據；","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"場景","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"系統絕對併發量上來了，分表難以根本上解決問題，並且還沒有明顯的業務歸屬來垂直分庫。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"分析","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"庫多了，io和cpu的壓力自然可以成倍緩解。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"水平分表","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/54/545335c10d86e4e06bc60c78c9b1e5b1.png","alt":null,"title":"","style":[{"key":"width","value":"50%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"概念","attrs":{}}]},{"type":"paragraph","attrs":{"indent":2,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"以字段爲依據，按照一定策略（hash、range等），將一個表中的數據拆分到多個表中","attrs":{}},{"type":"text","text":"。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"結果","attrs":{}}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"每個表的結構都一樣；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"每個表的數據都不一樣，沒有交集；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"所有表的並集是全量數據；","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"場景","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"系統絕對併發量並沒有上來，只是單表的數據量太多，影響了SQL效率，加重了CPU負擔，以至於成爲瓶頸。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"分析","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"表的數據量少了，單次SQL執行效率高，自然減輕了CPU的負擔。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"垂直分庫","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/9a/9a177bc1018ef153bb2e3d7b7e497eb7.png","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"概念","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"以表爲依據，按照業務歸屬不同，將不同的表拆分到不同的庫中。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"結果","attrs":{}}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"每個庫的結構都不一樣；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"每個庫的數據也不一樣，沒有交集；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"所有庫的並集是全量數據；","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"場景","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":" 系統絕對併發量上來了，並且可以抽象出單獨的業務模塊。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"分析","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":" 到這一步，基本上就可以服務化了。例如，隨着業務的發展一些公用的配置表、字典表等越來越多，這時可以將這些表拆到單獨的庫中，甚至可以服務化。再有，隨着業務的發展孵化出了一套業務模式，這時可以將相關的表拆到單獨的庫中，甚至可以服務化。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"垂直分表","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/3b/3b21dfd7147a08057b01b04b11e401d3.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"1、概念：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":" 以字段爲依據，按照字段的活躍性，將表中字段拆到不同的表（主表和擴展表）中。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"2、結果：","attrs":{}}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"每個表的結構都不一樣；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"每個表的數據也不一樣，一般來說，每個表的字段至少有一列交集，一般是主鍵，用於關聯數據；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"所有表的並集是全量數據；","attrs":{}}]}]}],"attrs":{}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"3、場景： ","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":" ","attrs":{}},{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"系統絕對併發量並沒有上來，表的記錄並不多，但是字段多，並且熱點數據和非熱點數據在一起，單行數據所需的存儲空間較大。以至於數據庫緩存的數據行減少，查詢時會去讀磁盤數據產生大量的隨機讀IO，產生IO瓶頸","attrs":{}},{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"4、分析：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":" 可以用列表頁和詳情頁來幫助理解。","attrs":{}},{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"垂直分表的拆分原則是將熱點數據（可能會冗餘經常一起查詢的數據）放在一起作爲主表，非熱點數據放在一起作爲擴展表。","attrs":{}},{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"這樣更多的熱點數據就能被緩存下來，進而減少了隨機讀IO。拆了之後，要想獲得全部數據就需要關聯兩個表來取數據。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":" 但記住，千萬別用join，因爲join不僅會增加CPU負擔並且會講兩個表耦合在一起（必須在一個數據庫實例上）。關聯數據，應該在業務Service層做文章，分別獲取主表和擴展表數據然後用關聯字段關聯得到全部數據。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"分庫分表工具","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","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":"1、sharding-sphere：jar，前身是sharding-jdbc；","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":"2、TDDL：jar，Taobao Distribute Data Layer；","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":"3、Mycat：中間件","attrs":{}},{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#F5222D","name":"red"}}],"text":"。","attrs":{}}]}],"attrs":{}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":"注：工具的利弊，請自行調研，官網和社區優先。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"分庫分表步驟","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" 根據容量（當前容量和增長量）評估分庫或分表個數 -> 選key（均勻）-> 分表規則（hash或range等）-> 執行（一般雙寫）-> 擴容問題（儘量減少數據的移動）。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"分庫分表問題","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"非partition key的查詢問題：（水平分庫分表，拆分策略爲常用的hash法）","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"除了partition key只有一個非partition key作爲條件查詢","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":5},"content":[{"type":"text","text":"映射法","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/d9/d9d97d9738bb725ad9a4ee855ccf23c3.png","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"heading","attrs":{"align":null,"level":5},"content":[{"type":"text","text":"基因法","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/ed/ed3c787e72c35add29897171125fd7aa.png","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"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":" 注：寫入時，基因法生成user_id，如圖。關於xbit基因，例如要分8張表，23=8，故x取3，即3bit基因。根據user_id查詢時可直接取模路由到對應的分庫或分表。根據user_name查詢時，先通過user_name_code生成函數生成user_name_code再對其取模路由到對應的分庫或分表。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"id生成常用","attrs":{}},{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"snowflake算法。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"端上除了partition key不止一個非partition key作爲條件查詢","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"映射法","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/d9/d9d97d9738bb725ad9a4ee855ccf23c3.png","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"冗餘法","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/b9/b9df137f486a70af4f29837ebd8e2f9a.jpeg","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"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","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":"注：按照order_id或buyer_id查詢時路由到db_o_buyer庫中，按照seller_id查詢時路由到db_o_seller庫中。","attrs":{}}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"後臺除了partition key還有各種非partition key組合條件查詢","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"NoSQL法","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/fb/fbeb4c80040ae942c745e6d6fff920b6.png","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"冗餘法","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/ed/edf226e4bd6865c22e2c433b3a18fa7b.png","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"非partition key跨庫跨表分頁查詢問題","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"（水平分庫分表，拆分策略爲常用的hash法）","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"注：用NoSQL法解決（ES等）。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"擴容問題（水平分庫分表，拆分策略爲常用的hash法）","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"水平擴容庫（升級從庫法)","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/97/9777dd752cc5bf7a1f4f2da218092fdd.jpeg","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":"注：擴容是成倍的。","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","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":"水平擴容表（雙寫遷移法）","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"第一步：（同步雙寫）應用配置雙寫，部署；","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"第二步：（同步雙寫）將老庫中的老數據複製到新庫中；","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"第三步：（同步雙寫）以老庫爲準校對新庫中的老數據；","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"第四步：（同步雙寫）應用去掉雙寫，部署；","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":"注：雙寫是通用方案。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"分庫分表總結","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" 分庫分表，首先得知道瓶頸在哪裏，然後才能合理地拆分（分庫還是分表？水平還是垂直？分幾個？）。且不可爲了分庫分表而拆分。","attrs":{}}]},{"type":"numberedlist","attrs":{"start":1,"normalizeStart":1},"content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":1,"align":null,"origin":null},"content":[{"type":"text","text":"選key很重要，既要考慮到拆分均勻，也要考慮到非partition key的查詢。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":2,"align":null,"origin":null},"content":[{"type":"text","text":"只要能滿足需求，拆分規則越簡單越好。","attrs":{}}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}}]}
🌏【架構師指南】總結分庫分表的實現方案
{"type":"doc","content":[{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"每日一句","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","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#40A9FF","name":"blue"}},{"type":"strong","attrs":{}}],"text":"愛一個人要適度，和愛人的距離也要適度。愛他在不遠不近的距離，他便會對你不離不棄。","attrs":{}}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"數據庫瓶頸","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#000000","name":"black"}}],"text":" ","attrs":{}},{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#000000","name":"black"}},{"type":"strong","attrs":{}}],"text":"不管是IO瓶頸，還是CPU瓶頸，最終都會導致數據庫的活躍連接數增加，進而逼近甚至達到數據庫可承載活躍連接數的閾值。在業務Service來看就是，可用數據庫連接少甚至無連接可用。接下來就可以想象了吧（併發量、吞吐量、崩潰）。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"IO瓶頸","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#000000","name":"black"}},{"type":"strong","attrs":{}}],"text":" 第一種：磁盤讀IO瓶頸，熱點數據太多，數據庫緩存放不下，每次查詢時會產生大量的IO，降低查詢速度 -> 分庫和垂直分表","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#000000","name":"black"}},{"type":"strong","attrs":{}}],"text":" 第二種：網絡IO瓶頸，請求的數據太多，網絡帶寬不夠 ->分庫","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"CPU瓶頸","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" ","attrs":{}},{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":" ","attrs":{}},{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#000000","name":"black"}},{"type":"strong","attrs":{}}],"text":"第一種：SQL問題，如SQL中包含join，group by，order by，非索引字段條件查詢等，增加CPU運算的操作 -> SQL優化，建立合適的索引，在業務Service層進行業務計算。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#000000","name":"black"}},{"type":"strong","attrs":{}}],"text":" 第二種：單表數據量太大，查詢時掃描的行太多，SQL效率低，CPU率先出現瓶頸 -> 水平分表。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"分庫分表","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"水平分庫","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/7e/7e3611be103dc7bf850abf51a57dc40c.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":true,"pastePass":true}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"概念","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":" 以字段爲依據，按照一定策略（hash、range等），將一個庫中的數據拆分到多個庫中。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"結果","attrs":{}}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"每個庫的結構都一樣；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"每個庫的數據都不一樣，沒有交集；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"所有庫的並集是全量數據；","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"場景","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"系統絕對併發量上來了，分表難以根本上解決問題，並且還沒有明顯的業務歸屬來垂直分庫。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"分析","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"庫多了，io和cpu的壓力自然可以成倍緩解。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"水平分表","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/54/545335c10d86e4e06bc60c78c9b1e5b1.png","alt":null,"title":"","style":[{"key":"width","value":"50%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"概念","attrs":{}}]},{"type":"paragraph","attrs":{"indent":2,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"以字段爲依據，按照一定策略（hash、range等），將一個表中的數據拆分到多個表中","attrs":{}},{"type":"text","text":"。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"結果","attrs":{}}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"每個表的結構都一樣；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"每個表的數據都不一樣，沒有交集；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"所有表的並集是全量數據；","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"場景","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"系統絕對併發量並沒有上來，只是單表的數據量太多，影響了SQL效率，加重了CPU負擔，以至於成爲瓶頸。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"分析","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"表的數據量少了，單次SQL執行效率高，自然減輕了CPU的負擔。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"垂直分庫","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/9a/9a177bc1018ef153bb2e3d7b7e497eb7.png","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"概念","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"以表爲依據，按照業務歸屬不同，將不同的表拆分到不同的庫中。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"結果","attrs":{}}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"每個庫的結構都不一樣；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"每個庫的數據也不一樣，沒有交集；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"所有庫的並集是全量數據；","attrs":{}}]}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"場景","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":" 系統絕對併發量上來了，並且可以抽象出單獨的業務模塊。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"分析","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":" 到這一步，基本上就可以服務化了。例如，隨着業務的發展一些公用的配置表、字典表等越來越多，這時可以將這些表拆到單獨的庫中，甚至可以服務化。再有，隨着業務的發展孵化出了一套業務模式，這時可以將相關的表拆到單獨的庫中，甚至可以服務化。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"垂直分表","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/3b/3b21dfd7147a08057b01b04b11e401d3.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"1、概念：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":" 以字段爲依據，按照字段的活躍性，將表中字段拆到不同的表（主表和擴展表）中。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"2、結果：","attrs":{}}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"每個表的結構都不一樣；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"每個表的數據也不一樣，一般來說，每個表的字段至少有一列交集，一般是主鍵，用於關聯數據；","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"所有表的並集是全量數據；","attrs":{}}]}]}],"attrs":{}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"3、場景： ","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":" ","attrs":{}},{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"系統絕對併發量並沒有上來，表的記錄並不多，但是字段多，並且熱點數據和非熱點數據在一起，單行數據所需的存儲空間較大。以至於數據庫緩存的數據行減少，查詢時會去讀磁盤數據產生大量的隨機讀IO，產生IO瓶頸","attrs":{}},{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"4、分析：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":" 可以用列表頁和詳情頁來幫助理解。","attrs":{}},{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"strong","attrs":{}}],"text":"垂直分表的拆分原則是將熱點數據（可能會冗餘經常一起查詢的數據）放在一起作爲主表，非熱點數據放在一起作爲擴展表。","attrs":{}},{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"這樣更多的熱點數據就能被緩存下來，進而減少了隨機讀IO。拆了之後，要想獲得全部數據就需要關聯兩個表來取數據。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":" 但記住，千萬別用join，因爲join不僅會增加CPU負擔並且會講兩個表耦合在一起（必須在一個數據庫實例上）。關聯數據，應該在業務Service層做文章，分別獲取主表和擴展表數據然後用關聯字段關聯得到全部數據。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"分庫分表工具","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","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":"1、sharding-sphere：jar，前身是sharding-jdbc；","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":"2、TDDL：jar，Taobao Distribute Data Layer；","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":"3、Mycat：中間件","attrs":{}},{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#F5222D","name":"red"}}],"text":"。","attrs":{}}]}],"attrs":{}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":"注：工具的利弊，請自行調研，官網和社區優先。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"分庫分表步驟","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" 根據容量（當前容量和增長量）評估分庫或分表個數 -> 選key（均勻）-> 分表規則（hash或range等）-> 執行（一般雙寫）-> 擴容問題（儘量減少數據的移動）。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"分庫分表問題","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"非partition key的查詢問題：（水平分庫分表，拆分策略爲常用的hash法）","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"除了partition key只有一個非partition key作爲條件查詢","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":5},"content":[{"type":"text","text":"映射法","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/d9/d9d97d9738bb725ad9a4ee855ccf23c3.png","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"heading","attrs":{"align":null,"level":5},"content":[{"type":"text","text":"基因法","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/ed/ed3c787e72c35add29897171125fd7aa.png","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"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":" 注：寫入時，基因法生成user_id，如圖。關於xbit基因，例如要分8張表，23=8，故x取3，即3bit基因。根據user_id查詢時可直接取模路由到對應的分庫或分表。根據user_name查詢時，先通過user_name_code生成函數生成user_name_code再對其取模路由到對應的分庫或分表。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"id生成常用","attrs":{}},{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"snowflake算法。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"端上除了partition key不止一個非partition key作爲條件查詢","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"映射法","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/d9/d9d97d9738bb725ad9a4ee855ccf23c3.png","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"冗餘法","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/b9/b9df137f486a70af4f29837ebd8e2f9a.jpeg","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"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","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":"注：按照order_id或buyer_id查詢時路由到db_o_buyer庫中，按照seller_id查詢時路由到db_o_seller庫中。","attrs":{}}]}],"attrs":{}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"後臺除了partition key還有各種非partition key組合條件查詢","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"NoSQL法","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/fb/fbeb4c80040ae942c745e6d6fff920b6.png","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"冗餘法","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/ed/edf226e4bd6865c22e2c433b3a18fa7b.png","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"非partition key跨庫跨表分頁查詢問題","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"（水平分庫分表，拆分策略爲常用的hash法）","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"注：用NoSQL法解決（ES等）。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"擴容問題（水平分庫分表，拆分策略爲常用的hash法）","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"水平擴容庫（升級從庫法)","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/97/9777dd752cc5bf7a1f4f2da218092fdd.jpeg","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"border"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":"注：擴容是成倍的。","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","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":"水平擴容表（雙寫遷移法）","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"第一步：（同步雙寫）應用配置雙寫，部署；","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"第二步：（同步雙寫）將老庫中的老數據複製到新庫中；","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"第三步：（同步雙寫）以老庫爲準校對新庫中的老數據；","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"第四步：（同步雙寫）應用去掉雙寫，部署；","attrs":{}}]},{"type":"paragraph","attrs":{"indent":1,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong","attrs":{}}],"text":"注：雙寫是通用方案。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"分庫分表總結","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" 分庫分表，首先得知道瓶頸在哪裏，然後才能合理地拆分（分庫還是分表？水平還是垂直？分幾個？）。且不可爲了分庫分表而拆分。","attrs":{}}]},{"type":"numberedlist","attrs":{"start":1,"normalizeStart":1},"content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":1,"align":null,"origin":null},"content":[{"type":"text","text":"選key很重要，既要考慮到拆分均勻，也要考慮到非partition key的查詢。","attrs":{}}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":1,"number":2,"align":null,"origin":null},"content":[{"type":"text","text":"只要能滿足需求，拆分規則越簡單越好。","attrs":{}}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}}]}
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