auto commit

notes/MySQL.md

@@ -1,226 +1,227 @@
 <!-- GFM-TOC -->
-* [<EFBFBD>洢<EFBFBD><EFBFBD><EFBFBD><EFBFBD>](#<23>洢<EFBFBD><E6B4A2><EFBFBD><EFBFBD>)
+* [存储引擎](#存储引擎)
     * [1. InnoDB](#1-innodb)
     * [2. MyISAM](#2-myisam)
-    * [3. InnoDB <EFBFBD><EFBFBD> MyISAM <EFBFBD>ıȽ<EFBFBD>](#3-innodb-<EFBFBD><EFBFBD>-myisam-<EFBFBD>ıȽ<EFBFBD>)
-* [<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>](#<23><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>)
-    * [1. <EFBFBD><EFBFBD><EFBFBD><EFBFBD>](#1-<EFBFBD><EFBFBD><EFBFBD><EFBFBD>)
-    * [2. <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>](#2-<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>)
-    * [3. <EFBFBD>ַ<EFBFBD><EFBFBD><EFBFBD>](#3-<EFBFBD>ַ<EFBFBD><EFBFBD><EFBFBD>)
-    * [4. ʱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>](#4-ʱ<><CAB1><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>)
-* [<EFBFBD><EFBFBD><EFBFBD><EFBFBD>](#<EFBFBD><EFBFBD><EFBFBD><EFBFBD>)
-    * [1. <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>](#1-<2D><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>)
-        * [1.1 B-Tree <EFBFBD><EFBFBD><EFBFBD><EFBFBD>](#11-b-tree-<EFBFBD><EFBFBD><EFBFBD><EFBFBD>)
-        * [1.2 <EFBFBD><EFBFBD>ϣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>](#12-<EFBFBD><EFBFBD>ϣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>)
-        * [1.3. <EFBFBD>ռ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݣ<EFBFBD>R-Tree<EFBFBD><EFBFBD>](#13-<EFBFBD>ռ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>r-tree)
-        * [1.4 ȫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>](#14-ȫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>)
-    * [2. <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ŵ<EFBFBD>](#2-<2D><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ŵ<EFBFBD>)
-    * [3. <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ż<EFBFBD>](#3-<2D><><EFBFBD><EFBFBD><EFBFBD>Ż<EFBFBD>)
-        * [3.1 <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>](#31-<2D><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>)
-        * [3.2 ǰ׺<EFBFBD><EFBFBD><EFBFBD><EFBFBD>](#32-ǰ׺<EFBFBD><EFBFBD><EFBFBD><EFBFBD>)
-        * [3.3 <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>](#33-<2D><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>)
-        * [3.4 <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>е<EFBFBD>˳<EFBFBD><EFBFBD>](#34-<2D><><EFBFBD><EFBFBD><EFBFBD>е<EFBFBD>˳<EFBFBD><CBB3>)
-        * [3.5 <EFBFBD>۴<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>](#35-<EFBFBD>۴<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>)
-        * [3.6 <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>](#36-<2D><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>)
-    * [4. B-Tree <EFBFBD><EFBFBD> B+Tree ԭ<EFBFBD><EFBFBD>](#4-b-tree-<EFBFBD><EFBFBD>-b+tree-ԭ<EFBFBD><EFBFBD>)
+    * [3. InnoDB 与 MyISAM 的比较](#3-innodb-与-myisam-的比较)
+* [数据类型](#数据类型)
+    * [1. 整型](#1-整型)
+    * [2. 浮点数](#2-浮点数)
+    * [3. 字符串](#3-字符串)
+    * [4. 时间和日期](#4-时间和日期)
+* [索引](#索引)
+    * [1. 索引分类](#1-索引分类)
+        * [1.1 B-Tree 索引](#11-b-tree-索引)
+        * [1.2 哈希索引](#12-哈希索引)
+        * [1.3. 空间索引数据（R-Tree）](#13-空间索引数据r-tree)
+        * [1.4 全文索引](#14-全文索引)
+    * [2. 索引的优点](#2-索引的优点)
+    * [3. 索引优化](#3-索引优化)
+        * [3.1 独立的列](#31-独立的列)
+        * [3.2 前缀索引](#32-前缀索引)
+        * [3.3 多列索引](#33-多列索引)
+        * [3.4 索引列的顺序](#34-索引列的顺序)
+        * [3.5 聚簇索引](#35-聚簇索引)
+        * [3.6 覆盖索引](#36-覆盖索引)
+    * [4. B-Tree 和 B+Tree 原理](#4-b-tree-和-b+tree-原理)
         * [4. 1 B-Tree](#4-1-b-tree)
         * [4.2 B+Tree](#42-b+tree)
-        * [4.3 <EFBFBD><EFBFBD><EFBFBD><EFBFBD>˳<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ָ<EFBFBD><EFBFBD><EFBFBD><EFBFBD> B+Tree](#43-<2D><><EFBFBD><EFBFBD>˳<EFBFBD><CBB3><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ָ<EFBFBD><D6B8><EFBFBD><EFBFBD>-b+tree)
-        * [4.4 Ϊʲôʹ<EFBFBD><EFBFBD> B-Tree <EFBFBD><EFBFBD> B+Tree](#44-Ϊʲôʹ<EFBFBD><EFBFBD>-b-tree-<EFBFBD><EFBFBD>-b+tree)
-* [<EFBFBD><EFBFBD>ѯ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ż<EFBFBD>](#<23><>ѯ<EFBFBD><D1AF><EFBFBD><EFBFBD><EFBFBD>Ż<EFBFBD>)
+        * [4.3 带有顺序访问指针的 B+Tree](#43-带有顺序访问指针的-b+tree)
+        * [4.4 为什么使用 B-Tree 和 B+Tree](#44-为什么使用-b-tree-和-b+tree)
+* [查询性能优化](#查询性能优化)
     * [1. Explain](#1-explain)
-    * [2. <EFBFBD><EFBFBD><EFBFBD>ٷ<EFBFBD><EFBFBD>ص<EFBFBD><EFBFBD><EFBFBD>](#2-<2D><><EFBFBD>ٷ<EFBFBD><D9B7>ص<EFBFBD><D8B5><EFBFBD>)
-    * [3. <EFBFBD><EFBFBD><EFBFBD>ٷ<EFBFBD><EFBFBD>ص<EFBFBD><EFBFBD><EFBFBD>](#3-<2D><><EFBFBD>ٷ<EFBFBD><D9B7>ص<EFBFBD><D8B5><EFBFBD>)
-    * [4. <EFBFBD><EFBFBD><EFBFBD>ִ<EFBFBD><EFBFBD><EFBFBD> DELETE <EFBFBD><EFBFBD> INSERT <EFBFBD><EFBFBD><EFBFBD><EFBFBD>](#4-<EFBFBD><EFBFBD><EFBFBD>ִ<EFBFBD><EFBFBD><EFBFBD>-delete-<EFBFBD><EFBFBD>-insert-<EFBFBD><EFBFBD><EFBFBD><EFBFBD>)
-* [<EFBFBD>ֿ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֱ<EFBFBD>](#<23>ֿ<EFBFBD><D6BF><EFBFBD><EFBFBD>ֱ<EFBFBD>)
-* [<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ת<EFBFBD>ƺ͹<EFBFBD><EFBFBD>ϻָ<EFBFBD>](#<23><><EFBFBD><EFBFBD>ת<EFBFBD>ƺ͹<C6BA><CDB9>ϻָ<CFBB>)
-    * [1. <EFBFBD><EFBFBD><EFBFBD><EFBFBD>ת<EFBFBD><EFBFBD>](#1-<2D><><EFBFBD><EFBFBD>ת<EFBFBD><D7AA>)
-    * [2. <EFBFBD><EFBFBD><EFBFBD>ϻָ<EFBFBD>](#2-<EFBFBD><EFBFBD><EFBFBD>ϻָ<EFBFBD>)
-* [<EFBFBD>ο<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>](#<23>ο<EFBFBD><CEBF><EFBFBD><EFBFBD><EFBFBD>)
+    * [2. 减少返回的列](#2-减少返回的列)
+    * [3. 减少返回的行](#3-减少返回的行)
+    * [4. 拆分大的 DELETE 或 INSERT 语句](#4-拆分大的-delete-或-insert-语句)
+* [分库与分表](#分库与分表)
+* [故障转移和故障恢复](#故障转移和故障恢复)
+    * [1. 故障转移](#1-故障转移)
+    * [2. 故障恢复](#2-故障恢复)
+* [参考资料](#参考资料)
 <!-- GFM-TOC -->
 
-# <20>洢<EFBFBD><E6B4A2><EFBFBD><EFBFBD>
+
+# 存储引擎
 
 ## 1. InnoDB
 
-InnoDB <EFBFBD><EFBFBD> MySQL <EFBFBD><EFBFBD>Ĭ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>棬ֻ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ҫ InnoDB <20><>֧<EFBFBD>ֵ<EFBFBD><D6B5><EFBFBD><EFBFBD><EFBFBD>ʱ<EFBFBD><CAB1><EFBFBD>ſ<EFBFBD><C5BF><EFBFBD>ʹ<EFBFBD><CAB9><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>洢<EFBFBD><E6B4A2><EFBFBD>档
+InnoDB 是 MySQL 的默认事务型引擎，只有在需要 InnoDB 不支持的特性时，才考虑使用其它存储引擎。
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD> MVCC <EFBFBD><EFBFBD>֧<EFBFBD>ָ߲<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ĸ<EFBFBD><EFBFBD><EFBFBD>׼<EFBFBD>ĸ<EFBFBD><EFBFBD>뼶<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ĭ<EFBFBD>ϼ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ǿ<EFBFBD><EFBFBD>ظ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+采用 MVCC 来支持高并发，并且实现了四个标准的隔离级别，默认级别是可重复读。
 
-<EFBFBD><EFBFBD><EFBFBD>ǻ<EFBFBD><EFBFBD>ھ۴<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ģ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>IJ<EFBFBD>ѯ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>кܸߵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+表是基于聚簇索引建立的，它对主键的查询性能有很高的提升。
 
-<EFBFBD>ڲ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>˺ܶ<EFBFBD><EFBFBD>Ż<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ӵ<EFBFBD><EFBFBD>̶<EFBFBD>ȡ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʱ<EFBFBD><EFBFBD><EFBFBD>õĿ<EFBFBD>Ԥ<EFBFBD><EFBFBD><EFBFBD>Զ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ܹ<EFBFBD><EFBFBD>Զ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڴ<EFBFBD><EFBFBD>д<EFBFBD><EFBFBD><EFBFBD> hash <20><><EFBFBD><EFBFBD><EFBFBD>Լ<EFBFBD><D4BC>ٶ<EFBFBD><D9B6><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ӧ<EFBFBD><D3A6>ϣ<EFBFBD><CFA3><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Լ<EFBFBD><D4BC>ܹ<EFBFBD><DCB9><EFBFBD><EFBFBD>ٲ<EFBFBD><D9B2><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>IJ<EFBFBD><C4B2>뻺<EFBFBD><EBBBBA><EFBFBD><EFBFBD><EFBFBD>ȡ<EFBFBD>
+内部做了很多优化，包括从磁盘读取数据时采用的可预测性读，能够自动在内存中创建 hash 索引以加速读操作的自适应哈希索引，以及能够加速插入操作的插入缓冲区等。
 
-ͨ<EFBFBD><EFBFBD>һЩ<EFBFBD><EFBFBD><EFBFBD>ƺ͹<EFBFBD><EFBFBD><EFBFBD>֧<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ȱ<EFBFBD><EFBFBD>ݡ<EFBFBD>
+通过一些机制和工具支持真正的热备份。
 
 ## 2. MyISAM
 
-MyISAM <EFBFBD>ṩ<EFBFBD>˴<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ԣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ȫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ѹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ռ亯<EFBFBD><EFBFBD><EFBFBD><EFBFBD>GIS<EFBFBD><EFBFBD><EFBFBD>ȡ<EFBFBD><EFBFBD><EFBFBD> MyISAM <20><>֧<EFBFBD><D6A7><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>м<EFBFBD><D0BC><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ұ<EFBFBD><D2B1><EFBFBD><EFBFBD><EFBFBD><EFBFBD>޷<EFBFBD><DEB7><EFBFBD>ȫ<EFBFBD>ָ<EFBFBD><D6B8><EFBFBD>
+MyISAM 提供了大量的特性，包括全文索引、压缩、空间函数（GIS）等。但 MyISAM 不支持事务和行级锁，而且奔溃后无法安全恢复。
 
-ֻ<EFBFBD>ܶ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ű<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>С<EFBFBD>
+只能对整张表加锁，而不是针对行。
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Զ<EFBFBD>ִ<EFBFBD>м<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>޸<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ǻ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ָ<EFBFBD><EFBFBD>Լ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ָ<EFBFBD><EFBFBD><EFBFBD>ͬ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ܵ<EFBFBD><EFBFBD><EFBFBD>һЩ<EFBFBD><EFBFBD><EFBFBD>ݶ<EFBFBD>ʧ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>޸<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ƿdz<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ġ<EFBFBD>
+可以手工或者自动执行检查和修复操作，但是和事务恢复以及奔溃恢复不同，可能导致一些数据丢失，而且修复操作是非常慢的。
 
-<EFBFBD><EFBFBD><EFBFBD>԰<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>̬<EFBFBD><EFBFBD><EFBFBD>߾<EFBFBD>̬<EFBFBD><EFBFBD><EFBFBD>С<EFBFBD>
+可以包含动态或者静态的行。
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ָ<EFBFBD><EFBFBD><EFBFBD><EFBFBD> DELAY_KEY_WRITE ѡ<EFBFBD><EFBFBD><EFBFBD>ÿ<EFBFBD><EFBFBD><EFBFBD>޸<EFBFBD>ִ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>޸ĵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>д<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>̣<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ǻ<EFBFBD>д<EFBFBD><EFBFBD><EFBFBD>ڴ<EFBFBD><EFBFBD>еļ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֻ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>߹رձ<EFBFBD><EFBFBD><EFBFBD>ʱ<EFBFBD><EFBFBD><EFBFBD>ŻὫ<EFBFBD><EFBFBD>Ӧ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>д<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>̡<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ַ<EFBFBD>ʽ<EFBFBD><EFBFBD><EFBFBD>Լ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>д<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ܣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݿ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>𻵣<EFBFBD><EFBFBD><EFBFBD>Ҫִ<EFBFBD><EFBFBD><EFBFBD>޸<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+如果指定了 DELAY_KEY_WRITE 选项，在每次修改执行完成时，不会立即将修改的索引数据写入磁盘，而是会写到内存中的键缓冲区，只有在清理键缓冲区或者关闭表的时候才会将对应的索引块写入磁盘。这种方式可以极大的提升写入性能，但是在数据库或者主机奔溃时会造成索引损坏，需要执行修复操作。
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڴ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ժ󣬲<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ٽ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>޸IJ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ô<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ı<EFBFBD><EFBFBD>ʺϲ<EFBFBD><EFBFBD><EFBFBD> MyISAM ѹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+如果表在创建并导入数据以后，不会再进行修改操作，那么这样的表适合采用 MyISAM 压缩表。
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֻ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>߱<EFBFBD><EFBFBD>Ƚ<EFBFBD>С<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>޸<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ȼ<EFBFBD><EFBFBD><EFBFBD>Լ<EFBFBD><EFBFBD><EFBFBD>ʹ<EFBFBD><EFBFBD> MyISAM<EFBFBD><EFBFBD>
+对于只读数据，或者表比较小、可以容忍修复操作，则依然可以继续使用 MyISAM。
 
-MyISAM <EFBFBD><EFBFBD><EFBFBD>Ƽ򵥣<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Խ<EFBFBD><EFBFBD>ܸ<EFBFBD>ʽ<EFBFBD>洢<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ijЩ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ܺܺá<EFBFBD>
+MyISAM 设计简单，数据以紧密格式存储，所以在某些场景下性能很好。
 
-## 3. InnoDB <EFBFBD><EFBFBD> MyISAM <EFBFBD>ıȽ<EFBFBD>
+## 3. InnoDB 与 MyISAM 的比较
 
-**<EFBFBD><EFBFBD><EFBFBD><EFBFBD>**
+**事务**
 
-InnoDB <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>͵ġ<EFBFBD>
+InnoDB 是事务型的。
 
-**<EFBFBD><EFBFBD><EFBFBD><EFBFBD>**
+**备份**
 
-InnoDB ֧<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ȱ<EFBFBD><EFBFBD>ݡ<EFBFBD>
+InnoDB 支持在线热备份。
 
-**<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ָ<EFBFBD>**
+**奔溃恢复**
 
-MyISAM <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>𻵵ĸ<EFBFBD><EFBFBD>ʱ<EFBFBD> InnoDB <20>ߺܶ࣬<DCB6><E0A3AC><EFBFBD>һָ<D2BB><D6B8><EFBFBD><EFBFBD>ٶ<EFBFBD>Ҳ<EFBFBD><D2B2><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+MyISAM 奔溃后发生损坏的概率比 InnoDB 高很多，而且恢复的速度也更慢。
 
-**<EFBFBD><EFBFBD><EFBFBD><EFBFBD>**
+**并发**
 
-MyISAM ֻ֧<EFBFBD>ֱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> InnoDB <20><>֧<EFBFBD><D6A7><EFBFBD>м<EFBFBD><D0BC><EFBFBD><EFBFBD><EFBFBD>
+MyISAM 只支持表级锁，而 InnoDB 还支持行级锁。
 
-**<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>**
+**其它特性**
 
-MyISAM ֧<EFBFBD><EFBFBD>ȫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ռ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+MyISAM 支持全文索引，地理空间索引；
 
-# <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+# 数据类型
 
-## 1. <EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+## 1. 整型
 
-TINYINT, SMALLINT, MEDIUMINT, INT, BIGINT <EFBFBD>ֱ<EFBFBD>ʹ<EFBFBD><EFBFBD> 8, 16, 24, 64 λ<EFBFBD>洢<EFBFBD>ռ䣬һ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ԽС<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Խ<EFBFBD>á<EFBFBD>
+TINYINT, SMALLINT, MEDIUMINT, INT, BIGINT 分别使用 8, 16, 24, 64 位存储空间，一般情况下越小的列越好。
 
-INT(11) <EFBFBD>е<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֻ<EFBFBD>ǹ涨<EFBFBD>˽<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʾ<EFBFBD>ַ<EFBFBD><EFBFBD>ĸ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڴ洢<EFBFBD>ͼ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>˵<EFBFBD><EFBFBD>û<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ġ<EFBFBD>
+INT(11) 中的数字只是规定了交互工具显示字符的个数，对于存储和计算来说是没有意义的。
 
-## 2. <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+## 2. 浮点数
 
-FLOAT <EFBFBD><EFBFBD> DOUBLE Ϊ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͣ<EFBFBD>DECIMAL Ϊ<>߾<EFBFBD><DFBE><EFBFBD>С<EFBFBD><D0A1><EFBFBD><EFBFBD><EFBFBD>͡<EFBFBD>CPU ԭ<><D4AD>֧<EFBFBD>ָ<EFBFBD><D6B8><EFBFBD><EFBFBD><EFBFBD><EFBFBD>㣬<EFBFBD><E3A3AC><EFBFBD>Dz<EFBFBD>֧<EFBFBD><D6A7> DECIMAl <20><><EFBFBD>͵ļ<CDB5><C4BC>㣬<EFBFBD><E3A3AC><EFBFBD><EFBFBD> DECIMAL <20>ļ<EFBFBD><C4BC><EFBFBD><EFBFBD>ȸ<EFBFBD><C8B8><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ҫ<EFBFBD><D2AA><EFBFBD>ߵĴ<DFB5><C4B4>ۡ<EFBFBD>
+FLOAT 和 DOUBLE 为浮点类型，DECIMAL 为高精度小数类型。CPU 原生支持浮点运算，但是不支持 DECIMAl 类型的计算，因此 DECIMAL 的计算比浮点类型需要更高的代价。
 
-FLOAT<EFBFBD><EFBFBD>DOUBLE <EFBFBD><EFBFBD> DECIMAL <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ָ<EFBFBD><EFBFBD><EFBFBD>п<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> DECIMAL(18, 9) <20><>ʾ<EFBFBD>ܹ<EFBFBD> 18 λ<><CEBB>ȡ 9 λ<>洢С<E6B4A2><D0A1><EFBFBD><EFBFBD><EFBFBD>֣<EFBFBD>ʣ<EFBFBD><CAA3> 9 λ<>洢<EFBFBD><E6B4A2><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>֡<EFBFBD>
+FLOAT、DOUBLE 和 DECIMAL 都可以指定列宽，例如 DECIMAL(18, 9) 表示总共 18 位，取 9 位存储小数部分，剩下 9 位存储整数部分。
 
-## 3. <EFBFBD>ַ<EFBFBD><EFBFBD><EFBFBD>
+## 3. 字符串
 
-<EFBFBD><EFBFBD>Ҫ<EFBFBD><EFBFBD> CHAR <EFBFBD><EFBFBD> VARCHAR <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͣ<EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD>Ƕ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ģ<EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD>DZ䳤<EFBFBD>ġ<EFBFBD>
+主要有 CHAR 和 VARCHAR 两种类型，一种是定长的，一种是变长的。
 
-VARCHAR <EFBFBD><EFBFBD><EFBFBD>ֱ䳤<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ܹ<EFBFBD><EFBFBD><EFBFBD>ʡ<EFBFBD>ռ䣬<EFBFBD><EFBFBD>Ϊֻ<EFBFBD><EFBFBD>Ҫ<EFBFBD>洢<EFBFBD><EFBFBD>Ҫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݡ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ִ<EFBFBD><EFBFBD> UPDATE ʱ<><CAB1><EFBFBD>ܻ<EFBFBD>ʹ<EFBFBD>б<EFBFBD><D0B1>ñ<EFBFBD>ԭ<EFBFBD><D4AD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ<EFBFBD><D2BB>ҳ<EFBFBD><D2B3><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ɵĴ<C9B5>Сʱ<D0A1><CAB1><EFBFBD><EFBFBD>Ҫִ<D2AA>ж<EFBFBD><D0B6><EFBFBD><EFBFBD>IJ<EFBFBD><C4B2><EFBFBD><EFBFBD><EFBFBD>MyISAM <20>Ὣ<EFBFBD>в<EFBFBD><D0B2>ɲ<EFBFBD>ͬ<EFBFBD><CDAC>Ƭ<EFBFBD>δ洢<CEB4><E6B4A2><EFBFBD><EFBFBD> InnoDB <20><><EFBFBD><EFBFBD>Ҫ<EFBFBD><D2AA><EFBFBD><EFBFBD>ҳ<EFBFBD><D2B3>ʹ<EFBFBD>зŽ<D0B7>ҳ<EFBFBD>ڡ<EFBFBD>
+VARCHAR 这种变长类型能够节省空间，因为只需要存储必要的内容。但是在执行 UPDATE 时可能会使行变得比原来长，当超出一个页所能容纳的大小时，就要执行额外的操作，MyISAM 会将行拆成不同的片段存储，而 InnoDB 则需要分裂页来使行放进页内。
 
-VARCHAR <EFBFBD>ᱣ<EFBFBD><EFBFBD><EFBFBD>ַ<EFBFBD><EFBFBD><EFBFBD>ĩβ<EFBFBD>Ŀո񣬶<EFBFBD> CHAR <20><>ɾ<EFBFBD><C9BE><EFBFBD><EFBFBD>
+VARCHAR 会保留字符串末尾的空格，而 CHAR 会删除。
 
-## 4. ʱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+## 4. 时间和日期
 
-MySQL <EFBFBD>ṩ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ƶ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͣ<EFBFBD>DATATIME <EFBFBD><EFBFBD> TIMESTAMP<EFBFBD><EFBFBD>
+MySQL 提供了两种相似的日期时间类型：DATATIME 和 TIMESTAMP。
 
 **DATATIME**
 
-<EFBFBD>ܹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> 1001 <EFBFBD>굽 9999 <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ں<EFBFBD>ʱ<EFBFBD>䣬<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ϊ<EFBFBD>룬ʹ<EFBFBD><EFBFBD> 8 <20>ֽڵĴ洢<C4B4>ռ䡣
+能够保存从 1001 年到 9999 年的日期和时间，精度为秒，使用 8 字节的存储空间。
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʱ<EFBFBD><EFBFBD><EFBFBD>޹ء<EFBFBD>
+它与时区无关。
 
-Ĭ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>£<EFBFBD>MySQL <20><>һ<EFBFBD>ֿ<EFBFBD><D6BF><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ġ<EFBFBD><C4A1><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ĸ<EFBFBD>ʽ<EFBFBD><CABD>ʾ DATATIME ֵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>硰2008-01016 22:37:08<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> ANSI <20><>׼<EFBFBD><D7BC><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ں<EFBFBD>ʱ<EFBFBD><CAB1><EFBFBD><EFBFBD>ʾ<EFBFBD><CABE><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+默认情况下，MySQL 以一种可排序的、无歧义的格式显示 DATATIME 值，例如“2008-01016 22:37:08”，这是 ANSI 标准定义的日期和时间表示方法。
 
 **TIMESTAMP**
 
-<EFBFBD><EFBFBD> UNIX ʱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͬ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> 1970 <EFBFBD><EFBFBD> 1 <EFBFBD><EFBFBD> 1 <EFBFBD><EFBFBD><EFBFBD><EFBFBD>ҹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʱ<EFBFBD>䣩<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʹ<EFBFBD><EFBFBD> 4 <20><><EFBFBD>ֽڣ<D6BD>ֻ<EFBFBD>ܱ<EFBFBD>ʾ<EFBFBD><CABE> 1970 <EFBFBD><EFBFBD> <20><> 2038 <EFBFBD>ꡣ
+和 UNIX 时间戳相同，保存从 1970 年 1 月 1 日午夜（格林威治时间）以来的秒数，使用 4 个字节，只能表示从 1970 年 到 2038 年。
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʱ<EFBFBD><EFBFBD><EFBFBD>йء<EFBFBD>
+它和时区有关。
 
-MySQL <EFBFBD>ṩ<EFBFBD><EFBFBD> FROM_UNIXTIME() <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> Unxi ʱ<><CAB1><EFBFBD><EFBFBD>ת<EFBFBD><D7AA>Ϊ<EFBFBD><CEAA><EFBFBD>ڣ<EFBFBD><DAA3><EFBFBD><EFBFBD>ṩ<EFBFBD><E1B9A9> UNIX_TIMESTAMP() <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ת<EFBFBD><D7AA>Ϊ Unix ʱ<><CAB1><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+MySQL 提供了 FROM_UNIXTIME() 函数把 Unxi 时间戳转换为日期，并提供了 UNIX_TIMESTAMP() 函数把日期转换为 Unix 时间戳。
 
-Ĭ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>£<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʱû<EFBFBD><EFBFBD>ָ<EFBFBD><EFBFBD> TIMESTAMP <20>е<EFBFBD>ֵ<EFBFBD><D6B5><EFBFBD>Ὣ<EFBFBD><E1BDAB><EFBFBD><EFBFBD>ֵ<EFBFBD><D6B5><EFBFBD><EFBFBD>Ϊ<EFBFBD><CEAA>ǰʱ<C7B0>䡣
+默认情况下，如果插入时没有指定 TIMESTAMP 列的值，会将这个值设置为当前时间。
 
-Ӧ<EFBFBD>þ<EFBFBD><EFBFBD><EFBFBD>ʹ<EFBFBD><EFBFBD> TIMESTAMP<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ϊ<EFBFBD><EFBFBD><EFBFBD><EFBFBD> DATETIME <EFBFBD>ռ<EFBFBD>Ч<EFBFBD>ʸ<EFBFBD><EFBFBD>ߡ<EFBFBD>
+应该尽量使用 TIMESTAMP，因为它比 DATETIME 空间效率更高。
 
-# <EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+# 索引
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڴ洢<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʵ<EFBFBD>ֵģ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڷ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʵ<EFBFBD>ֵģ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Բ<EFBFBD>ͬ<EFBFBD>洢<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>в<EFBFBD>ͬ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͺ<EFBFBD>ʵ<EFBFBD>֡<EFBFBD>
+索引是在存储引擎层实现的，而不是在服务器层实现的，所以不同存储引擎具有不同的索引类型和实现。
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ܹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>׽<EFBFBD><EFBFBD><EFBFBD>ѯ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+索引能够轻易将查询性能提升几个数量级。
 
-<EFBFBD><EFBFBD><EFBFBD>ڷdz<EFBFBD>С<EFBFBD>ı<EFBFBD><EFBFBD><EFBFBD><EFBFBD>󲿷<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>¼򵥵<EFBFBD>ȫ<EFBFBD><EFBFBD>ɨ<EFBFBD><EFBFBD><EFBFBD>Ƚ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ч<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>е<EFBFBD><EFBFBD><EFBFBD><EFBFBD>͵ı<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͷdz<EFBFBD><EFBFBD><EFBFBD>Ч<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ƕ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ش<EFBFBD><EFBFBD>͵ı<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ĵ<EFBFBD><EFBFBD>۽<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>֮<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>£<EFBFBD><EFBFBD><EFBFBD>Ҫ<EFBFBD>õ<EFBFBD>һ<EFBFBD>ּ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֳ<EFBFBD><EFBFBD><EFBFBD>Ҫ<EFBFBD><EFBFBD>ѯ<EFBFBD><EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>¼һ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>¼<EFBFBD><EFBFBD>ƥ<EFBFBD>䣬<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʹ<EFBFBD>÷<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+对于非常小的表、大部分情况下简单的全表扫描比建立索引更高效。对于中到大型的表，索引就非常有效。但是对于特大型的表，建立和使用索引的代价将会随之增长。这种情况下，需要用到一种技术可以直接区分出需要查询的一组数据，而不是一条记录一条记录地匹配，例如可以使用分区技术。
 
-## 1. <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+## 1. 索引分类
 
-### 1.1 B-Tree <EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+### 1.1 B-Tree 索引
 
-B-Tree <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ǵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> MySQL <20>洢<EFBFBD><E6B4A2><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ĭ<EFBFBD><C4AC><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>͡<EFBFBD>
+B-Tree 索引是大多数 MySQL 存储引擎的默认索引类型。
 
-<EFBFBD><EFBFBD>Ϊ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ҫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ȫ<EFBFBD><EFBFBD>ɨ<EFBFBD>裬ֻ<EFBFBD><EFBFBD>Ҫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ɣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>˲<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ٶȿ<EFBFBD><EFBFBD>ܶࡣ
+因为不再需要进行全表扫描，只需要对树进行搜索即可，因此查找速度快很多。
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ָ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ϊ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>У<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>й<EFBFBD>ͬ<EFBFBD><EFBFBD><EFBFBD>ɼ<EFBFBD><EFBFBD><EFBFBD>B-Tree <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ȫ<EFBFBD><C8AB>ֵ<EFBFBD><D6B5><EFBFBD><EFBFBD>ֵ<EFBFBD><D6B5>Χ<EFBFBD>ͼ<EFBFBD>ǰ׺<C7B0><D7BA><EFBFBD>ң<EFBFBD><D2A3><EFBFBD><EFBFBD>м<EFBFBD>ǰ׺<C7B0><D7BA><EFBFBD><EFBFBD>ֻ<EFBFBD><D6BB><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ǰ׺<C7B0><D7BA><EFBFBD>ҡ<EFBFBD>
+可以指定多个列作为索引列，多个索引列共同组成键。B-Tree 索引适用于全键值、键值范围和键前缀查找，其中键前缀查找只适用于最左前缀查找。
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڲ<EFBFBD><EFBFBD>ң<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͷ<EFBFBD><EFBFBD>顣
+除了用于查找，还可以用于排序和分组。
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ǰ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>е<EFBFBD>˳<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>в<EFBFBD><EFBFBD>ң<EFBFBD><EFBFBD><EFBFBD><EFBFBD>޷<EFBFBD>ʹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+如果不是按照索引列的顺序进行查找，则无法使用索引。
 
-### 1.2 <EFBFBD><EFBFBD>ϣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+### 1.2 哈希索引
 
-<EFBFBD><EFBFBD><EFBFBD>ڹ<EFBFBD>ϣ<EFBFBD><EFBFBD>ʵ<EFBFBD>֣<EFBFBD><EFBFBD>ŵ<EFBFBD><EFBFBD>Dz<EFBFBD><EFBFBD>ҷdz<EFBFBD><EFBFBD>졣
+基于哈希表实现，优点是查找非常快。
 
-<EFBFBD><EFBFBD> MySQL <EFBFBD><EFBFBD>ֻ<EFBFBD><EFBFBD> Memory <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʽ֧<EFBFBD>ֹ<EFBFBD>ϣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+在 MySQL 中只有 Memory 引擎显式支持哈希索引。
 
-InnoDB <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ĺ<EFBFBD><EFBFBD>ܽС<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ӧ<EFBFBD><EFBFBD>ϣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ij<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֵ<EFBFBD><EFBFBD>ʹ<EFBFBD>õķdz<EFBFBD>Ƶ<EFBFBD><EFBFBD>ʱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> B-Tree <20><><EFBFBD><EFBFBD>֮<EFBFBD><D6AE><EFBFBD>ٴ<EFBFBD><D9B4><EFBFBD>һ<EFBFBD><D2BB><EFBFBD><EFBFBD>ϣ<EFBFBD><CFA3><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> B-Tree <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>й<EFBFBD>ϣ<EFBFBD><CFA3><EFBFBD><EFBFBD><EFBFBD><EFBFBD>һЩ<D2BB>ŵ㣬<C5B5><E3A3AC><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ٵĹ<D9B5>ϣ<EFBFBD><CFA3><EFBFBD>ҡ<EFBFBD>
+InnoDB 引擎有一个特殊的功能叫“自适应哈希索引”，当某个索引值被使用的非常频繁时，会在 B-Tree 索引之上再创建一个哈希索引，这样就让 B-Tree 索引具有哈希索引的一些优点，比如快速的哈希查找。
 
-<EFBFBD><EFBFBD><EFBFBD>ƣ<EFBFBD><EFBFBD><EFBFBD>ϣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֻ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ϣֵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ָ<EFBFBD>룬<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>洢<EFBFBD>ֶ<EFBFBD>ֵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Բ<EFBFBD><EFBFBD><EFBFBD>ʹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>е<EFBFBD>ֵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ȡ<EFBFBD>С<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڴ<EFBFBD><EFBFBD>е<EFBFBD><EFBFBD>е<EFBFBD><EFBFBD>ٶȺܿ죬<EFBFBD><EFBFBD><EFBFBD>Դ󲿷<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ӱ<EFBFBD>첢<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ԣ<EFBFBD><EFBFBD>޷<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڷ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֻ֧<EFBFBD>־<EFBFBD>ȷ<EFBFBD><EFBFBD><EFBFBD>ң<EFBFBD><EFBFBD>޷<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڲ<EFBFBD><EFBFBD>ֲ<EFBFBD><EFBFBD>Һͷ<EFBFBD>Χ<EFBFBD><EFBFBD><EFBFBD>ң<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ϣ<EFBFBD><EFBFBD>ͻ<EFBFBD>ܶ࣬<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ٶȻ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ú<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+限制：哈希索引只包含哈希值和行指针，而不存储字段值，所以不能使用索引中的值来避免读取行。不过，访问内存中的行的速度很快，所以大部分情况下这一点对性能影响并不明显；无法用于分组与排序；只支持精确查找，无法用于部分查找和范围查找；如果哈希冲突很多，查找速度会变得很慢。
 
-### 1.3. <EFBFBD>ռ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݣ<EFBFBD>R-Tree<EFBFBD><EFBFBD>
+### 1.3. 空间索引数据（R-Tree）
 
-MyISAM <EFBFBD>洢<EFBFBD><EFBFBD><EFBFBD><EFBFBD>֧<EFBFBD>ֿռ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݴ洢<EFBFBD><EFBFBD>
+MyISAM 存储引擎支持空间索引，可以用于地理数据存储。
 
-<EFBFBD>ռ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ά<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ч<EFBFBD><EFBFBD>ʹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ά<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ϲ<EFBFBD>ѯ<EFBFBD><EFBFBD>
+空间索引会从所有维度来索引数据，可以有效地使用任意维度来进行组合查询。
 
-### 1.4 ȫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+### 1.4 全文索引
 
-MyISAM <EFBFBD>洢<EFBFBD><EFBFBD><EFBFBD><EFBFBD>֧<EFBFBD><EFBFBD>ȫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڲ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ı<EFBFBD><EFBFBD>еĹؼ<EFBFBD><EFBFBD>ʣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֱ<EFBFBD>ӱȽ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>е<EFBFBD>ֵ<EFBFBD><EFBFBD>
+MyISAM 存储引擎支持全文索引，用于查找文本中的关键词，而不是直接比较索引中的值。
 
-ʹ<EFBFBD><EFBFBD> MATCH AGAINST<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͨ<EFBFBD><EFBFBD> WHERE<EFBFBD><EFBFBD>
+使用 MATCH AGAINST，而不是普通的 WHERE。
 
-## 2. <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ŵ<EFBFBD>
+## 2. 索引的优点
 
-- <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>˷<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ҫɨ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+- 大大减少了服务器需要扫描的数据量；
 
-- <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+- 帮助服务器避免进行排序和创建临时表；
 
-- <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> I/O <20><>Ϊ˳<CEAA><CBB3> I/O<EFBFBD><EFBFBD>
+- 将随机 I/O 变为顺序 I/O。
 
-## 3. <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ż<EFBFBD>
+## 3. 索引优化
 
-### 3.1 <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+### 3.1 独立的列
 
-<EFBFBD>ڽ<EFBFBD><EFBFBD>в<EFBFBD>ѯʱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>в<EFBFBD><EFBFBD><EFBFBD><EFBFBD>DZ<EFBFBD><EFBFBD><EFBFBD>ʽ<EFBFBD><EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD>֣<EFBFBD>Ҳ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ǻ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>IJ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>޷<EFBFBD>ʹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+在进行查询时，索引列不能是表达式的一部分，也不能是函数的参数，否则无法使用索引。
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>IJ<EFBFBD>ѯ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʹ<EFBFBD><EFBFBD> actor_id <EFBFBD>е<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+例如下面的查询不能使用 actor_id 列的索引：
 
 ```sql
 SELECT actor_id FROM sakila.actor WHERE actor_id + 1 = 5;
 ```
 
-### 3.2 ǰ׺<EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+### 3.2 前缀索引
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD> BLOB<EFBFBD><EFBFBD>TEXT <EFBFBD><EFBFBD> VARCHAR <EFBFBD><EFBFBD><EFBFBD>͵<EFBFBD><EFBFBD>У<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʹ<EFBFBD><EFBFBD>ǰ׺<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֻ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʼ<EFBFBD>IJ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ַ<EFBFBD><EFBFBD><EFBFBD>
+对于 BLOB、TEXT 和 VARCHAR 类型的列，必须使用前缀索引，只索引开始的部分字符。
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ǰ׺<EFBFBD><EFBFBD><EFBFBD>ȵ<EFBFBD>ѡȡ<EFBFBD><EFBFBD>Ҫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD> **<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ѡ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>** <20><>ȷ<EFBFBD><C8B7><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ظ<EFBFBD><D8B8><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֵ<EFBFBD>ͼ<EFBFBD>¼<EFBFBD><C2BC><EFBFBD><EFBFBD><EFBFBD>ı<EFBFBD>ֵ<EFBFBD><D6B5>ѡ<EFBFBD><D1A1><EFBFBD><EFBFBD>Խ<EFBFBD>ߣ<EFBFBD><DFA3><EFBFBD>ѯЧ<D1AF><D0A7>ҲԽ<D2B2>ߡ<EFBFBD><DFA1><EFBFBD><EFBFBD><EFBFBD>ֵΪ 1 <20><><EFBFBD><EFBFBD>ʱÿ<CAB1><C3BF><EFBFBD><EFBFBD>¼<EFBFBD><C2BC><EFBFBD><EFBFBD>Ψһ<CEA8><D2BB><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ӧ<EFBFBD><D3A6>
+对于前缀长度的选取需要根据 **索引选择性** 来确定：不重复的索引值和记录总数的比值。选择性越高，查询效率也越高。最大值为 1 ，此时每个记录都有唯一的索引与其对应。
 
-### 3.3 <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+### 3.3 多列索引
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ҫʹ<EFBFBD>ö<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ϊ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>в<EFBFBD>ѯʱ<EFBFBD><EFBFBD>ʹ<EFBFBD>ö<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʹ<EFBFBD>ö<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ܸ<EFBFBD><EFBFBD>á<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>У<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ð<EFBFBD> actor_id <EFBFBD><EFBFBD> file_id <EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ϊ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+在需要使用多个列作为条件进行查询时，使用多列索引比使用多个单列索引性能更好。例如下面的语句中，最好把 actor_id 和 file_id 设置为多列索引。
 
 ```sql
 SELECT file_id, actor_ id FROM sakila.film_actor
 WhERE actor_id = 1 OR film_id = 1;
 ```
 
-### 3.4 <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>е<EFBFBD>˳<EFBFBD><EFBFBD>
+### 3.4 索引列的顺序
 
-<EFBFBD><EFBFBD>ѡ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ǿ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>з<EFBFBD><EFBFBD><EFBFBD>ǰ<EFBFBD>棬<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʾ<EFBFBD>Ľ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> customer_id <EFBFBD><EFBFBD>ѡ<EFBFBD><EFBFBD><EFBFBD>Ա<EFBFBD> staff_id <EFBFBD><EFBFBD><EFBFBD>ߣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ð<EFBFBD> customer_id <20>з<EFBFBD><D0B7>ڶ<EFBFBD><DAB6><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ǰ<EFBFBD>档
+让选择性最强的索引列放在前面，例如下面显示的结果中 customer_id 的选择性比 staff_id 更高，因此最好把 customer_id 列放在多列索引的前面。
 
 ```sql
 SELECT COUNT(DISTINCT staff_id)/COUNT(*) AS staff_id_selectivity,
@@ -235,107 +236,107 @@ customer_id_selectivity: 0.0373
                COUNT(*): 16049
 ```
 
-### 3.5 <EFBFBD>۴<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+### 3.5 聚簇索引
 
-![](https://github.com/CyC2018/InterviewNotes/blob/master/pics/b9e9ae8c-e216-4c01-b267-a50dbeb98fa4.jpg)
+![](https://github.com/CyC2018/InterviewNotes/blob/master/pics//b9e9ae8c-e216-4c01-b267-a50dbeb98fa4.jpg)
 
-<EFBFBD>۴<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݴ洢<EFBFBD><EFBFBD>ʽ<EFBFBD><EFBFBD>
+聚簇索引并不是一种索引类型，而是一种数据存储方式。
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD>۴ء<EFBFBD><EFBFBD><EFBFBD>ʾ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>к<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڵļ<EFBFBD>ֵ<EFBFBD><EFBFBD><EFBFBD>ܵش洢<EFBFBD><EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>InnoDB <20>ľ۴<C4BE><DBB4><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>д<EFBFBD><D0B4><EFBFBD><EFBFBD><EFBFBD> B-Tree <20><>Ҷ<EFBFBD><D2B6>ҳ<EFBFBD>С<EFBFBD>
+术语“聚簇”表示数据行和相邻的键值紧密地存储在一起，InnoDB 的聚簇索引的数据行存放在 B-Tree 的叶子页中。
 
-<EFBFBD><EFBFBD>Ϊ<EFBFBD>޷<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>д<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͬ<EFBFBD>ĵط<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֻ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD>۴<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+因为无法把数据行存放在两个不同的地方，所以一个表只能有一个聚簇索引。
 
-**<EFBFBD>ŵ<EFBFBD>**
+**优点**
 
-1. <EFBFBD><EFBFBD><EFBFBD>԰<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ<EFBFBD>𣬼<EFBFBD><EFBFBD><EFBFBD> I/O <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
-2. <EFBFBD><EFBFBD>Ϊ<EFBFBD><EFBFBD><EFBFBD>ݱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> B-Tree <20>У<EFBFBD><D0A3><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݷ<EFBFBD><DDB7>ʸ<EFBFBD><CAB8>졣
+1. 可以把相关数据保存在一起，减少 I/O 操作；
+2. 因为数据保存在 B-Tree 中，因此数据访问更快。
 
-**ȱ<EFBFBD><EFBFBD>**
+**缺点**
 
-1. <EFBFBD>۴<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>޶<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> I/O <20>ܼ<EFBFBD><DCBC><EFBFBD>Ӧ<EFBFBD>õ<EFBFBD><C3B5><EFBFBD><EFBFBD>ܣ<EFBFBD><DCA3><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ȫ<EFBFBD><C8AB><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڴ棬<DAB4><E6A3AC>û<EFBFBD><C3BB>Ҫ<EFBFBD>þ۴<C3BE><DBB4><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
-2. <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ٶ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڲ<EFBFBD><EFBFBD><EFBFBD>˳<EFBFBD>򣬰<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>˳<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ġ<EFBFBD>
-3. <EFBFBD><EFBFBD><EFBFBD>²<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ۺܸߣ<EFBFBD><EFBFBD><EFBFBD>Ϊÿ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>µ<EFBFBD><EFBFBD>ж<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ƶ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>µ<EFBFBD>λ<EFBFBD>á<EFBFBD>
-4. <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>뵽ij<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ҳ<EFBFBD>У<EFBFBD><EFBFBD>洢<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ὣ<EFBFBD><EFBFBD>ҳ<EFBFBD><EFBFBD><EFBFBD>ѳ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ҳ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ɸ<EFBFBD><EFBFBD>У<EFBFBD>ҳ<EFBFBD><EFBFBD><EFBFBD>ѻᵼ<EFBFBD>±<EFBFBD>ռ<EFBFBD>ø<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ĵ<EFBFBD><EFBFBD>̿ռ䡣
-5. <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>бȽ<EFBFBD>ϡ<EFBFBD>裬<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ҳ<EFBFBD><EFBFBD><EFBFBD>ѵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݴ洢<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʱ<EFBFBD><EFBFBD><EFBFBD>۴<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ܵ<EFBFBD><EFBFBD><EFBFBD>ȫ<EFBFBD><EFBFBD>ɨ<EFBFBD><EFBFBD><EFBFBD>ٶȱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+1. 聚簇索引最大限度提高了 I/O 密集型应用的性能，但是如果数据全部放在内存，就没必要用聚簇索引。
+2. 插入速度严重依赖于插入顺序，按主键的顺序插入是最快的。
+3. 更新操作代价很高，因为每个被更新的行都会移动到新的位置。
+4. 当插入到某个已满的页中，存储引擎会将该页分裂成两个页面来容纳该行，页分裂会导致表占用更多的磁盘空间。
+5. 如果行比较稀疏，或者由于页分裂导致数据存储不连续时，聚簇索引可能导致全表扫描速度变慢。
 
-### 3.6 <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+### 3.6 覆盖索引
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ҫ<EFBFBD><EFBFBD>ѯ<EFBFBD><EFBFBD><EFBFBD>ֶε<EFBFBD>ֵ<EFBFBD><EFBFBD>
+索引包含所有需要查询的字段的值。
 
-## 4. B-Tree <EFBFBD><EFBFBD> B+Tree ԭ<EFBFBD><EFBFBD>
+## 4. B-Tree 和 B+Tree 原理
 
 ### 4. 1 B-Tree
 
-![](https://github.com/CyC2018/InterviewNotes/blob/master/pics/5ed71283-a070-4b21-85ae-f2cbfd6ba6e1.jpg)
+![](https://github.com/CyC2018/InterviewNotes/blob/master/pics//5ed71283-a070-4b21-85ae-f2cbfd6ba6e1.jpg)
 
-Ϊ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> B-Tree<65><65><EFBFBD><EFBFBD><EFBFBD>ȶ<EFBFBD><C8B6><EFBFBD>һ<EFBFBD><D2BB><EFBFBD><EFBFBD><EFBFBD>ݼ<EFBFBD>¼Ϊһ<CEAA><D2BB><EFBFBD><EFBFBD>Ԫ<EFBFBD><D4AA> [key, data]<EFBFBD><EFBFBD>key Ϊ<EFBFBD><EFBFBD>¼<EFBFBD>ļ<EFBFBD><EFBFBD><EFBFBD>data Ϊ<><CEAA><EFBFBD>ݼ<EFBFBD>¼<EFBFBD><C2BC> key <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݡ<EFBFBD>
+为了描述 B-Tree，首先定义一条数据记录为一个二元组 [key, data]，key 为记录的键，data 为数据记录除 key 外的数据。
 
-B-Tree <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݽṹ<EFBFBD><EFBFBD>
+B-Tree 是满足下列条件的数据结构：
 
-- <EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ҷ<EFBFBD>ڵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͬ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ȣ<EFBFBD>Ҳ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>˵ B-Tree <20><>ƽ<EFBFBD><C6BD><EFBFBD>ģ<EFBFBD>
-- һ<EFBFBD><EFBFBD><EFBFBD>ڵ<EFBFBD><EFBFBD>е<EFBFBD> key <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ҷǵݼ<C7B5><DDBC><EFBFBD><EFBFBD>У<EFBFBD>
-- <EFBFBD><EFBFBD><EFBFBD><EFBFBD>ij<EFBFBD><EFBFBD>ָ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> key <20>ֱ<EFBFBD><D6B1><EFBFBD> key<sub>i</sub> <EFBFBD><EFBFBD> key<sub>i+1</sub><EFBFBD><EFBFBD><EFBFBD>Ҳ<EFBFBD>Ϊ null<6C><6C><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ָ<EFBFBD><D6B8>ָ<EFBFBD><D6B8><EFBFBD>ڵ<EFBFBD><DAB5><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> key <EFBFBD><EFBFBD><EFBFBD><EFBFBD> key<sub>i</sub> <EFBFBD><EFBFBD>С<EFBFBD><EFBFBD> key<sub>i+1</sub><EFBFBD><EFBFBD>
+- 所有叶节点具有相同的深度，也就是说 B-Tree 是平衡的；
+- 一个节点中的 key 从左到右非递减排列；
+- 如果某个指针的左右相邻 key 分别是 key<sub>i</sub> 和 key<sub>i+1</sub>，且不为 null，则该指针指向节点的所有 key 大于 key<sub>i</sub> 且小于 key<sub>i+1</sub>。
 
-<EFBFBD><EFBFBD> B-Tree <EFBFBD>а<EFBFBD> key <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݵ<EFBFBD><EFBFBD>㷨<EFBFBD>dz<EFBFBD>ֱ<EFBFBD>ۣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ȴӸ<EFBFBD><EFBFBD>ڵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ж<EFBFBD><EFBFBD>ֲ<EFBFBD><EFBFBD>ң<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ҵ<EFBFBD><EFBFBD>򷵻ض<EFBFBD>Ӧ<EFBFBD>ڵ<EFBFBD><EFBFBD><EFBFBD> data<74><61><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ӧ<EFBFBD><D3A6><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ָ<EFBFBD><D6B8>ָ<EFBFBD><D6B8><EFBFBD>Ľڵ<C4BD><DAB5>ݹ<EFBFBD><DDB9><EFBFBD><EFBFBD>в<EFBFBD><D0B2>ң<EFBFBD>ֱ<EFBFBD><D6B1><EFBFBD>ҵ<EFBFBD><D2B5>ڵ<EFBFBD><DAB5><EFBFBD><EFBFBD>ҵ<EFBFBD> null ָ<>룬ǰ<EBA3AC>߲<EFBFBD><DFB2>ҳɹ<D2B3><C9B9><EFBFBD><EFBFBD><EFBFBD><EFBFBD>߲<EFBFBD><DFB2><EFBFBD>ʧ<EFBFBD>ܡ<EFBFBD>
+在 B-Tree 中按 key 检索数据的算法非常直观：首先从根节点进行二分查找，如果找到则返回对应节点的 data，否则对相应区间的指针指向的节点递归进行查找，直到找到节点或找到 null 指针，前者查找成功，后者查找失败。
 
-<EFBFBD><EFBFBD><EFBFBD>ڲ<EFBFBD><EFBFBD><EFBFBD>ɾ<EFBFBD><EFBFBD><EFBFBD>µ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݼ<EFBFBD>¼<EFBFBD><EFBFBD><EFBFBD>ƻ<EFBFBD> B-Tree <20><><EFBFBD><EFBFBD><EFBFBD>ʣ<EFBFBD><CAA3><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڲ<EFBFBD><DAB2><EFBFBD>ɾ<EFBFBD><C9BE>ʱ<EFBFBD><CAB1><EFBFBD><EFBFBD>Ҫ<EFBFBD><D2AA><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ<EFBFBD><D2BB><EFBFBD><EFBFBD><EFBFBD>ѡ<EFBFBD><D1A1>ϲ<EFBFBD><CFB2><EFBFBD>ת<EFBFBD>ƵȲ<C6B5><C8B2><EFBFBD><EFBFBD>Ա<EFBFBD><D4B1><EFBFBD> B-Tree <EFBFBD><EFBFBD><EFBFBD>ʡ<EFBFBD>
+由于插入删除新的数据记录会破坏 B-Tree 的性质，因此在插入删除时，需要对树进行一个分裂、合并、转移等操作以保持 B-Tree 性质。
 
 ### 4.2 B+Tree
 
-![](https://github.com/CyC2018/InterviewNotes/blob/master/pics/63cd5b50-d6d8-4df6-8912-ef4a1dd5ba13.jpg)
+![](https://github.com/CyC2018/InterviewNotes/blob/master/pics//63cd5b50-d6d8-4df6-8912-ef4a1dd5ba13.jpg)
 
-<EFBFBD><EFBFBD> B-Tree <EFBFBD><EFBFBD><EFBFBD>ȣ<EFBFBD>B+Tree <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>²<EFBFBD>ͬ<EFBFBD>㣺
+与 B-Tree 相比，B+Tree 有以下不同点：
 
-- ÿ<EFBFBD><EFBFBD><EFBFBD>ڵ<EFBFBD><EFBFBD><EFBFBD>ָ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ϊ 2d <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD> 2d+1<EFBFBD><EFBFBD>
-- <EFBFBD>ڽڵ㲻<EFBFBD>洢 data<74><61>ֻ<EFBFBD>洢 key<65><79>Ҷ<EFBFBD>ӽڵ㲻<DAB5>洢ָ<E6B4A2>롣
+- 每个节点的指针上限为 2d 而不是 2d+1；
+- 内节点不存储 data，只存储 key，叶子节点不存储指针。
 
-### 4.3 <EFBFBD><EFBFBD><EFBFBD><EFBFBD>˳<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ָ<EFBFBD><EFBFBD><EFBFBD><EFBFBD> B+Tree
+### 4.3 带有顺序访问指针的 B+Tree
 
-![](https://github.com/CyC2018/InterviewNotes/blob/master/pics/1ee5f0a5-b8df-43b9-95ab-c516c54ec797.jpg)
+![](https://github.com/CyC2018/InterviewNotes/blob/master/pics//1ee5f0a5-b8df-43b9-95ab-c516c54ec797.jpg)
 
-һ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݿ<EFBFBD>ϵͳ<EFBFBD><EFBFBD><EFBFBD>ļ<EFBFBD>ϵͳ<EFBFBD><EFBFBD>ʹ<EFBFBD>õ<EFBFBD> B+Tree <20>ṹ<EFBFBD><E1B9B9><EFBFBD>ھ<EFBFBD><DABE><EFBFBD> B+Tree <20><><EFBFBD><EFBFBD><EFBFBD>Ͻ<EFBFBD><CFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ż<EFBFBD><C5BB><EFBFBD><EFBFBD><EFBFBD>Ҷ<EFBFBD>ӽڵ<D3BD><DAB5><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>˳<EFBFBD><CBB3><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ָ<EFBFBD>룬<EFBFBD><EBA3AC><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ż<EFBFBD><C5BB><EFBFBD>Ŀ<EFBFBD><C4BF><EFBFBD><EFBFBD>Ϊ<EFBFBD><CEAA><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʵ<EFBFBD><CAB5><EFBFBD><EFBFBD>ܡ<EFBFBD>
+一般在数据库系统或文件系统中使用的 B+Tree 结构都在经典 B+Tree 基础上进行了优化，在叶子节点增加了顺序访问指针，做这个优化的目的是为了提高区间访问的性能。
 
-### 4.4 Ϊʲôʹ<EFBFBD><EFBFBD> B-Tree <EFBFBD><EFBFBD> B+Tree
+### 4.4 为什么使用 B-Tree 和 B+Tree
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݽṹҲ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ļ<EFBFBD>ϵͳ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݿ<EFBFBD>ϵͳ<EFBFBD>ձ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> B-/+Tree <20><>Ϊ<EFBFBD><CEAA><EFBFBD><EFBFBD><EFBFBD>ṹ<EFBFBD><E1B9B9>
+红黑树等数据结构也可以用来实现索引，但是文件系统及数据库系统普遍采用 B-/+Tree 作为索引结构。
 
-ҳ<EFBFBD>Ǽ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>洢<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>߼<EFBFBD><EFBFBD>飬Ӳ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ϵͳ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʹ<EFBFBD><EFBFBD>̴洢<EFBFBD><EFBFBD><EFBFBD>ָ<EFBFBD>Ϊ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ĵ<EFBFBD>С<EFBFBD><EFBFBD><EFBFBD>ȵĿ飬ÿ<EFBFBD><EFBFBD><EFBFBD>洢<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ϊһҳ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ϵͳ<EFBFBD>У<EFBFBD>ҳ<EFBFBD>ô<EFBFBD>Сͨ<EFBFBD><EFBFBD>Ϊ 4k<34><6B><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʹ<EFBFBD><CDB4><EFBFBD><EFBFBD><EFBFBD>ҳΪ<D2B3><CEAA>λ<EFBFBD><CEBB><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݡ<EFBFBD>
+页是计算机管理存储器的逻辑块，硬件及操作系统往往将主存和磁盘存储区分割为连续的大小相等的块，每个存储块称为一页（在许多操作系统中，页得大小通常为 4k），主存和磁盘以页为单位交换数据。
 
-һ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>˵<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ҳ<EFBFBD>ܴ󣬲<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ȫ<EFBFBD><EFBFBD><EFBFBD>洢<EFBFBD><EFBFBD><EFBFBD>ڴ<EFBFBD><EFBFBD>У<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ļ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʽ<EFBFBD>洢<EFBFBD>Ĵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ϡ<EFBFBD>Ϊ<EFBFBD>˼<EFBFBD><EFBFBD>ٴ<EFBFBD><EFBFBD><EFBFBD> I/O<><4F><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ϸ<EFBFBD><CFB8><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ȡ<EFBFBD><C8A1><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ÿ<EFBFBD>ζ<EFBFBD><CEB6><EFBFBD>Ԥ<EFBFBD><D4A4><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ǽ<EFBFBD><C7BC><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ѧ<EFBFBD><D1A7><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ľֲ<C4BE><D6B2><EFBFBD>ԭ<EFBFBD><D4AD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ<EFBFBD><D2BB><EFBFBD><EFBFBD><EFBFBD>ݱ<EFBFBD><DDB1>õ<EFBFBD>ʱ<EFBFBD><CAB1><EFBFBD>丽<EFBFBD><E4B8BD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ҳͨ<D2B2><CDA8><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ϱ<EFBFBD>ʹ<EFBFBD>á<EFBFBD><C3A1><EFBFBD><EFBFBD>ݿ<EFBFBD>ϵͳ<CFB5><CDB3><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>˴<EFBFBD><CBB4><EFBFBD>Ԥ<EFBFBD><D4A4>ԭ<EFBFBD><D4AD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ<EFBFBD><D2BB><EFBFBD>ڵ<EFBFBD><DAB5>Ĵ<EFBFBD>С<EFBFBD><D0A1>Ϊ<EFBFBD><CEAA><EFBFBD><EFBFBD>һ<EFBFBD><D2BB>ҳ<EFBFBD><D2B3><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ÿ<EFBFBD><C3BF><EFBFBD>ڵ<EFBFBD>ֻ<EFBFBD><D6BB>Ҫһ<D2AA><D2BB> I/O <20>Ϳ<EFBFBD><CDBF><EFBFBD><EFBFBD><EFBFBD>ȫ<EFBFBD><C8AB><EFBFBD>롣B-Tree <20><>һ<EFBFBD>μ<EFBFBD><CEBC><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ҫ h-1 <EFBFBD><EFBFBD> I/O<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڵ㳣פ<EFBFBD>ڴ棩<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ӷ<EFBFBD>Ϊ O(h)=O(logdN)<29><>һ<EFBFBD><D2BB>ʵ<EFBFBD><CAB5>Ӧ<EFBFBD><D3A6><EFBFBD>У<EFBFBD><D0A3><EFBFBD><EFBFBD><EFBFBD> d <20>Ƿdz<C7B7><C7B3><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>֣<EFBFBD>ͨ<EFBFBD><CDA8><EFBFBD><EFBFBD><EFBFBD><EFBFBD> 100<30><30><EFBFBD><EFBFBD><EFBFBD><EFBFBD> h <20>dz<EFBFBD>С<EFBFBD><D0A1>ͨ<EFBFBD><CDA8><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> 3<><33><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֽṹ<D6BD><E1B9B9>h <20><><EFBFBD><EFBFBD>Ҫ<EFBFBD><D2AA><EFBFBD>Ķࡣ<C4B6><E0A1A3><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>߼<EFBFBD><DFBC>Ϻܽ<CFBA><DCBD>Ľڵ㣨<DAB5><E3A3A8><EFBFBD>ӣ<EFBFBD><D3A3><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ͽ<EFBFBD><CFBF>ܺ<EFBFBD>Զ<EFBFBD><D4B6><EFBFBD>޷<EFBFBD><DEB7><EFBFBD><EFBFBD>þֲ<C3BE><D6B2>ԣ<EFBFBD>Ч<EFBFBD><D0A7><EFBFBD><EFBFBD><EFBFBD>Ա<EFBFBD> B-Tree <EFBFBD><EFBFBD><EFBFBD>ܶࡣ
+一般来说，索引本身也很大，不可能全部存储在内存中，因此索引往往以索引文件的形式存储的磁盘上。为了减少磁盘 I/O，磁盘往往不是严格按需读取，而是每次都会预读。这样做的理论依据是计算机科学中著名的局部性原理：当一个数据被用到时，其附近的数据也通常会马上被使用。数据库系统的设计者巧妙利用了磁盘预读原理，将一个节点的大小设为等于一个页，这样每个节点只需要一次 I/O 就可以完全载入。B-Tree 中一次检索最多需要 h-1 次 I/O（根节点常驻内存），渐进复杂度为 O(h)=O(logdN)。一般实际应用中，出度 d 是非常大的数字，通常超过 100，因此 h 非常小（通常不超过 3）。而红黑树这种结构，h 明显要深的多。并且于逻辑上很近的节点（父子）物理上可能很远，无法利用局部性，效率明显比 B-Tree 差很多。
 
-B+Tree <EFBFBD><EFBFBD><EFBFBD>ʺ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ԭ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڽڵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> d <20>йء<D0B9><D8A1><EFBFBD><EFBFBD><EFBFBD> B+Tree <20>ڽڵ<DABD>ȥ<EFBFBD><C8A5><EFBFBD><EFBFBD> data <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>˿<EFBFBD><CBBF><EFBFBD>ӵ<EFBFBD>и<EFBFBD><D0B8><EFBFBD><EFBFBD>ij<EFBFBD><C4B3>ȣ<EFBFBD>ӵ<EFBFBD>и<EFBFBD><D0B8>õ<EFBFBD><C3B5><EFBFBD><EFBFBD>ܡ<EFBFBD>
+B+Tree 更适合外存索引，原因和内节点出度 d 有关。由于 B+Tree 内节点去掉了 data 域，因此可以拥有更大的出度，拥有更好的性能。
 
-# <EFBFBD><EFBFBD>ѯ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ż<EFBFBD>
+# 查询性能优化
 
 ## 1. Explain
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> SQL <20><><EFBFBD>䣬<EFBFBD><E4A3AC><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>бȽ<D0B1><C8BD><EFBFBD>Ҫ<EFBFBD><D2AA><EFBFBD>ֶ<EFBFBD><D6B6>У<EFBFBD>
+用来分析 SQL 语句，分析结果中比较重要的字段有：
 
-- select_type : <EFBFBD><EFBFBD>ѯ<EFBFBD><EFBFBD><EFBFBD>ͣ<EFBFBD><EFBFBD>м򵥲<EFBFBD>ѯ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ϲ<EFBFBD>ѯ<EFBFBD><EFBFBD><EFBFBD>Ӳ<EFBFBD>ѯ
+- select_type : 查询类型，有简单查询、联合查询和子查询
 
-- key : ʹ<EFBFBD>õ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+- key : 使用的索引
 
-- rows : ɨ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+- rows : 扫描的行数
 
-## 2. <EFBFBD><EFBFBD><EFBFBD>ٷ<EFBFBD><EFBFBD>ص<EFBFBD><EFBFBD><EFBFBD>
+## 2. 减少返回的列
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ѯ<EFBFBD><EFBFBD>Ҫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ϊ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>˹<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>˷<EFBFBD><EFBFBD>ʹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>֮<EFBFBD>⣬Ҳ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>С<EFBFBD>
+慢查询主要是因为访问了过多数据，除了访问过多行之外，也包括访问过多列。
 
-<EFBFBD><EFBFBD><EFBFBD>ò<EFBFBD>Ҫʹ<EFBFBD><EFBFBD> SELECT * <EFBFBD><EFBFBD><EFBFBD>䣬Ҫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ҫѡ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ѯ<EFBFBD><EFBFBD><EFBFBD>С<EFBFBD>
+最好不要使用 SELECT * 语句，要根据需要选择查询的列。
 
-## 3. <EFBFBD><EFBFBD><EFBFBD>ٷ<EFBFBD><EFBFBD>ص<EFBFBD><EFBFBD><EFBFBD>
+## 3. 减少返回的行
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʹ<EFBFBD><EFBFBD> LIMIT <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ȡ<EFBFBD><C8A1><EFBFBD><EFBFBD>Ҫ<EFBFBD><D2AA><EFBFBD><EFBFBD>Щ<EFBFBD>С<EFBFBD>
+最好使用 LIMIT 语句来取出想要的那些行。
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Խ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ȫ<EFBFBD><EFBFBD>ɨ<EFBFBD>衣<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>䣬<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ҫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ȫ<EFBFBD><EFBFBD>ɨ<EFBFBD>裬<EFBFBD><EFBFBD>ʹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֻ<EFBFBD><EFBFBD>Ҫɨ<EFBFBD>輸<EFBFBD>м<EFBFBD>¼<EFBFBD><EFBFBD><EFBFBD>ɣ<EFBFBD>ʹ<EFBFBD><EFBFBD> Explain <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͨ<EFBFBD><CDA8><EFBFBD>۲<EFBFBD> rows <20>ֶ<EFBFBD><D6B6><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֲ<EFBFBD><D6B2>졣
+还可以建立索引来减少条件语句的全表扫描。例如对于下面的语句，不适用索引的情况下需要进行全表扫描，而使用索引只需要扫描几行记录即可，使用 Explain 语句可以通过观察 rows 字段来看出这种差异。
 
 ```sql
 SELECT * FROM sakila.film_actor WHERE film_id = 1;
 ```
 
-## 4. <EFBFBD><EFBFBD><EFBFBD>ִ<EFBFBD><EFBFBD><EFBFBD> DELETE <EFBFBD><EFBFBD> INSERT <EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+## 4. 拆分大的 DELETE 或 INSERT 语句
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ִ<EFBFBD>еĻ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ס<EFBFBD>ܶ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݡ<EFBFBD>ռ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>־<EFBFBD><EFBFBD><EFBFBD>ľ<EFBFBD>ϵͳ<EFBFBD><EFBFBD>Դ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ܶ<EFBFBD>С<EFBFBD>ĵ<EFBFBD><EFBFBD><EFBFBD>Ҫ<EFBFBD>IJ<EFBFBD>ѯ<EFBFBD><EFBFBD>
+如果一次性执行的话，可能一次锁住很多数据、占满整个事务日志、耗尽系统资源、阻塞很多小的但重要的查询。
 
 ```sql
 DELEFT FROM messages WHERE create < DATE_SUB(NOW(), INTERVAL 3 MONTH);
@@ -348,74 +349,74 @@ do {
 } while rows_affected > 0
 ```
 
-# <EFBFBD>ֿ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֱ<EFBFBD>
+# 分库与分表
 
-**1. <EFBFBD>ֱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>IJ<EFBFBD>ͬ**
+**1. 分表与分区的不同**
 
-<EFBFBD>ֱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ǽ<EFBFBD>һ<EFBFBD>ű<EFBFBD><EFBFBD>ֳɶ<EFBFBD><EFBFBD><EFBFBD>С<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ЩС<EFBFBD><EFBFBD>ӵ<EFBFBD>в<EFBFBD>ͬ<EFBFBD>ı<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ǽ<EFBFBD>һ<EFBFBD>ű<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݷ<EFBFBD>Ϊ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>飬<EFBFBD><EFBFBD>Щ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Դ洢<EFBFBD><EFBFBD>ͬһ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ϣ<EFBFBD>Ҳ<EFBFBD><EFBFBD><EFBFBD>Դ洢<EFBFBD>ڲ<EFBFBD>ͬ<EFBFBD>Ĵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ϣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ַ<EFBFBD>ʽ<EFBFBD>±<EFBFBD><EFBFBD><EFBFBD>Ȼֻ<EFBFBD><EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+分表，就是讲一张表分成多个小表，这些小表拥有不同的表名；而分区是将一张表的数据分为多个区块，这些区块可以存储在同一个磁盘上，也可以存储在不同的磁盘上，这种方式下表仍然只有一个。
 
-**2. ʹ<EFBFBD>÷ֿ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֱ<EFBFBD><EFBFBD><EFBFBD>ԭ<EFBFBD><EFBFBD>**
+**2. 使用分库与分表的原因**
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ҵ<EFBFBD><EFBFBD><EFBFBD>ķ<EFBFBD>չ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݿ<EFBFBD><EFBFBD>еı<EFBFBD><EFBFBD><EFBFBD>Խ<EFBFBD><EFBFBD>Խ<EFBFBD>࣬<EFBFBD><EFBFBD><EFBFBD>ұ<EFBFBD><EFBFBD>е<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ҳ<EFBFBD><EFBFBD>Խ<EFBFBD><EFBFBD>Խ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ô<EFBFBD><EFBFBD>д<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ŀ<EFBFBD><EFBFBD><EFBFBD>Ҳ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+随着时间和业务的发展，数据库中的表会越来越多，并且表中的数据量也会越来越大，那么读写操作的开销也会随着增大。
 
-**3. <EFBFBD><EFBFBD>ֱ<EFBFBD>з<EFBFBD>**
+**3. 垂直切分**
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ģ<EFBFBD>顢<EFBFBD><EFBFBD>ϵ<EFBFBD><EFBFBD><EFBFBD>г̶Ȼ<EFBFBD><EFBFBD>ֳ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>𵽲<EFBFBD>ͬ<EFBFBD>Ŀ<EFBFBD><EFBFBD>ϡ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>磬<EFBFBD><EFBFBD><EFBFBD>ǻὨ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ʒ<EFBFBD><EFBFBD><EFBFBD>ݿ<EFBFBD> payDB<44><42><EFBFBD>û<EFBFBD><C3BB><EFBFBD><EFBFBD>ݿ<EFBFBD> userDB <20>ȣ<EFBFBD><C8A3>ֱ<EFBFBD><D6B1><EFBFBD><EFBFBD><EFBFBD><EFBFBD>洢<EFBFBD><E6B4A2>Ŀ<EFBFBD><C4BF><EFBFBD><EFBFBD>Ʒ<EFBFBD>йصı<D8B5><C4B1><EFBFBD><EFBFBD><EFBFBD><EFBFBD>û<EFBFBD><C3BB>йصı<D8B5><C4B1><EFBFBD>
+将表按功能模块、关系密切程度划分出来，部署到不同的库上。例如，我们会建立商品数据库 payDB、用户数据库 userDB 等，分别用来存储项目与商品有关的表和与用户有关的表。
 
-**4. ˮƽ<EFBFBD>з<EFBFBD>**
+**4. 水平切分**
 
-<EFBFBD>ѱ<EFBFBD><EFBFBD>е<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݰ<EFBFBD><EFBFBD><EFBFBD>ij<EFBFBD>ֹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>洢<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ṹ<EFBFBD><EFBFBD>ͬ<EFBFBD>ı<EFBFBD><EFBFBD>У<EFBFBD><EFBFBD><EFBFBD><EFBFBD>簴 id <20><>ɢ<EFBFBD><C9A2>ֵ<EFBFBD><D6B5><EFBFBD>Ա<EFBFBD><D4B1>Ƚ<EFBFBD><C8BD>л<EFBFBD><D0BB>֣<EFBFBD>
+把表中的数据按照某种规则存储到多个结构相同的表中，例如按 id 的散列值、性别等进行划分，
 
-**5. <EFBFBD><EFBFBD>ֱ<EFBFBD>з<EFBFBD><EFBFBD><EFBFBD>ˮƽ<EFBFBD>зֵ<EFBFBD>ѡ<EFBFBD><EFBFBD>**
+**5. 垂直切分与水平切分的选择**
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݿ<EFBFBD><EFBFBD>еı<EFBFBD>̫<EFBFBD>࣬<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ŀ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ҵ<EFBFBD><EFBFBD><EFBFBD>߼<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ô<EFBFBD><EFBFBD>ֱ<EFBFBD>з<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ѡ<EFBFBD><EFBFBD>
+如果数据库中的表太多，并且项目各项业务逻辑清晰，那么垂直切分是首选。
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݿ<EFBFBD><EFBFBD>ı<EFBFBD><EFBFBD><EFBFBD><EFBFBD>࣬<EFBFBD><EFBFBD><EFBFBD>ǵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ܴ<EFBFBD><EFBFBD><EFBFBD>Ӧ<EFBFBD><EFBFBD>ѡ<EFBFBD><EFBFBD>ˮƽ<EFBFBD>з֡<EFBFBD>
+如果数据库的表不多，但是单表的数据量很大，应该选择水平切分。
 
-**6. ˮƽ<EFBFBD>зֵ<EFBFBD>ʵ<EFBFBD>ַ<EFBFBD>ʽ**
+**6. 水平切分的实现方式**
 
-<EFBFBD><EFBFBD><EFBFBD>򵥵<EFBFBD><EFBFBD><EFBFBD>ʹ<EFBFBD><EFBFBD> merge <EFBFBD>洢<EFBFBD><EFBFBD><EFBFBD>档
+最简单的是使用 merge 存储引擎。
 
-**7. <EFBFBD>ֿ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>**
+**7. 分库与分表存在的问题**
 
-(1) <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+(1) 事务问题
 
-<EFBFBD><EFBFBD>ִ<EFBFBD>зֿ<EFBFBD><EFBFBD>ֱ<EFBFBD>֮<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݴ洢<EFBFBD><EFBFBD><EFBFBD>˲<EFBFBD>ͬ<EFBFBD>Ŀ<EFBFBD><EFBFBD>ϣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݿ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ѡ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݿⱾ<EFBFBD><EFBFBD><EFBFBD>ķֲ<EFBFBD>ʽ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ȥִ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>񣬽<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>߰<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ܴ<EFBFBD><EFBFBD>ۣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ӧ<EFBFBD>ó<EFBFBD><EFBFBD><EFBFBD>ȥЭ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ƣ<EFBFBD><EFBFBD>γɳ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>߼<EFBFBD><EFBFBD>ϵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֻ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ɱ<EFBFBD><EFBFBD>̷<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ĸ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+在执行分库分表之后，由于数据存储到了不同的库上，数据库事务管理出现了困难。如果依赖数据库本身的分布式事务管理功能去执行事务，将付出高昂的性能代价；如果由应用程序去协助控制，形成程序逻辑上的事务，又会造成编程方面的负担。
 
-(2) <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+(2) 跨库跨表连接问题
 
-<EFBFBD><EFBFBD>ִ<EFBFBD><EFBFBD><EFBFBD>˷ֿ<EFBFBD><EFBFBD>ֱ<EFBFBD>֮<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ա<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ὣԭ<EFBFBD><EFBFBD><EFBFBD>߼<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ժ<EFBFBD>ǿ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݻ<EFBFBD><EFBFBD>ֵ<EFBFBD><EFBFBD><EFBFBD>ͬ<EFBFBD>ı<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͬ<EFBFBD>Ŀ<EFBFBD><EFBFBD>ϡ<EFBFBD><EFBFBD><EFBFBD>ʱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ӳ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ܵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ƣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>޷<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>λ<EFBFBD>ڲ<EFBFBD>ͬ<EFBFBD>ֿ<EFBFBD><EFBFBD>ı<EFBFBD><EFBFBD><EFBFBD>Ҳ<EFBFBD>޷<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ӷֱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ȳ<EFBFBD>ͬ<EFBFBD>ı<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ԭ<EFBFBD><EFBFBD>ֻ<EFBFBD><EFBFBD>Ҫһ<EFBFBD>β<EFBFBD>ѯ<EFBFBD><EFBFBD><EFBFBD>ܹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ɵ<EFBFBD>ҵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ҫ<EFBFBD><EFBFBD><EFBFBD>ж<EFBFBD><EFBFBD>β<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ɡ<EFBFBD>
+在执行了分库分表之后，难以避免会将原本逻辑关联性很强的数据划分到不同的表、不同的库上。这时，表的连接操作将受到限制，我们无法连接位于不同分库的表，也无法连接分表粒度不同的表，导致原本只需要一次查询就能够完成的业务需要进行多次才能完成。
 
-# <EFBFBD><EFBFBD><EFBFBD><EFBFBD>ת<EFBFBD>ƺ͹<EFBFBD><EFBFBD>ϻָ<EFBFBD>
+# 故障转移和故障恢复
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ת<EFBFBD><EFBFBD>Ҳ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>л<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֹ<EFBFBD><EFBFBD><EFBFBD>ʱ<EFBFBD><EFBFBD><EFBFBD>л<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>⣬ʹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ϊ<EFBFBD><EFBFBD><EFBFBD>⡣<EFBFBD><EFBFBD><EFBFBD>ϻָ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>˼<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ǵӹ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>лָ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ұ<EFBFBD>֤<EFBFBD><EFBFBD><EFBFBD>ݵ<EFBFBD><EFBFBD><EFBFBD>ȷ<EFBFBD>ԡ<EFBFBD>
+故障转移也叫做切换，当主库出现故障时就切换到备库，使备库成为主库。故障恢复顾名思义就是从故障中恢复过来，并且保证数据的正确性。
 
-## 1. <EFBFBD><EFBFBD><EFBFBD><EFBFBD>ת<EFBFBD><EFBFBD>
+## 1. 故障转移
 
-**1.1 <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>л<EFBFBD><EFBFBD><EFBFBD>ɫ**
+**1.1 提升备库或切换角色**
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ̨<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ϊ<EFBFBD><EFBFBD><EFBFBD>⣬<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>-<2D><><EFBFBD><EFBFBD><EFBFBD>ƽṹ<C6BD>е<EFBFBD><D0B5><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͱ<EFBFBD><CDB1><EFBFBD><EFBFBD><EFBFBD>ɫ<EFBFBD><C9AB>
+提升一台备库为主库，或者在一个主-主复制结构中调整主动和被动角色。
 
-**1.2 <EFBFBD><EFBFBD><EFBFBD><EFBFBD> IP <20><>ַ<EFBFBD><D6B7> IP <EFBFBD>й<EFBFBD>**
+**1.2 虚拟 IP 地址和 IP 托管**
 
-Ϊ MySQL ʵ<EFBFBD><EFBFBD>ָ<EFBFBD><EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD>߼<EFBFBD> IP <20><>ַ<EFBFBD><D6B7><EFBFBD><EFBFBD> MySQL ʵ<><CAB5>ʧЧʱ<D0A7><CAB1><EFBFBD><EFBFBD><EFBFBD>Խ<EFBFBD> IP <20><>ַת<D6B7>Ƶ<EFBFBD><C6B5><EFBFBD>һ̨ MySQL <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ϡ<EFBFBD>
+为 MySQL 实例指定一个逻辑 IP 地址，当 MySQL 实例失效时，可以将 IP 地址转移到另一台 MySQL 服务器上。
 
-**1.3 <EFBFBD>м<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>**
+**1.3 中间件解决方案**
 
-ͨ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>·<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʹ<EFBFBD>õķ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ϡ<EFBFBD>
+通过代理，可以路由流量到可以使用的服务器上。
 
-![](https://github.com/CyC2018/InterviewNotes/blob/master/pics/fabd5fa0-b75e-48d0-9e2c-31471945ceb9.jpg)
+![](https://github.com/CyC2018/InterviewNotes/blob/master/pics//fabd5fa0-b75e-48d0-9e2c-31471945ceb9.jpg)
 
-**1.4 <EFBFBD><EFBFBD>Ӧ<EFBFBD><EFBFBD><EFBFBD>д<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ת<EFBFBD><EFBFBD>**
+**1.4 在应用中处理故障转移**
 
-<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ת<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ϵ<EFBFBD>Ӧ<EFBFBD><EFBFBD><EFBFBD>п<EFBFBD><EFBFBD>ܵ<EFBFBD><EFBFBD><EFBFBD>Ӧ<EFBFBD>ñ<EFBFBD><EFBFBD><EFBFBD>̫<EFBFBD><EFBFBD><EFBFBD><EFBFBD>׾<EFBFBD><EFBFBD>
+将故障转移整合到应用中可能导致应用变得太过笨拙。
 
-## 2. <EFBFBD><EFBFBD><EFBFBD>ϻָ<EFBFBD>
+## 2. 故障恢复
 
 
-# <EFBFBD>ο<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
+# 参考资料
 
-- <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> MySQL
-- [MySQL <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݽṹ<EFBFBD><EFBFBD><EFBFBD>㷨ԭ<EFBFBD><EFBFBD> ](http://blog.codinglabs.org/articles/theory-of-mysql-index.html)
-- [MySQL <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ż<EFBFBD>ȫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD> ](http://www.runoob.com/w3cnote/mysql-index.html)
-- [20+ <EFBFBD><EFBFBD> MySQL <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ż<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ѿ<EFBFBD><EFBFBD><EFBFBD> ](https://www.jfox.info/20-tiao-mysql-xing-nen-you-hua-de-zui-jia-jing-yan.html)
+- 高性能 MySQL
+- [MySQL 索引背后的数据结构及算法原理 ](http://blog.codinglabs.org/articles/theory-of-mysql-index.html)
+- [MySQL 索引优化全攻略 ](http://www.runoob.com/w3cnote/mysql-index.html)
+- [20+ 条 MySQL 性能优化的最佳经验 ](https://www.jfox.info/20-tiao-mysql-xing-nen-you-hua-de-zui-jia-jing-yan.html)