【数据库】Mysql中的索引与失效场景
1、前言
MySQL中数据表设计合理的索引对提高性能很有帮助。使用索引可以快速地定位表中的某条记录，从而提高数据库查询的速度，提高数据库的性能。
大多数情况下都（默认）采用B+ 树来构建索引。只是空间列类型的索引使R- 树，并且MEMORY 表还支持hash 索引。其实，用不用索引最终都是优化器说了算。 执行优化器是基于cost开销(CostBaseOptimizer) ，它不是基于规则（ Rule-BasedOptimizer），也不是基于语义。另外 SQL 语句是否使用索引，跟数据库版本、数据量、数据选择度都有关系。但是，还是有一些明显的规则可以去判别是否索引失效，比如：最左匹配原则、数据类型转换、反向判断等。
1.1、索引类型
MySQL支持多种类型的索引结构，包括B+tree、Hash、R-tree和Full-text等。
B+tree索引：MySQL中最常用的一种索引结构。B+tree是一种平衡树，每个节点的度数在一个范围之内，并且每个叶子节点都在同一层上。B+tree索引可以在极短的时间内进行查找、排序和插入等操作，适合于较小的索引和范围查询。
Hash索引：一种使用哈希表实现的索引结构。哈希索引在存储数据时将key和值映射到一个哈希表中的位置，查询时可以直接通过计算key的哈希值来快速访问数据。哈希索引适合等值查询（即查询操作中通过“=”来查询数据）。
R-tree索引：一种用于存储和查询基于位置的数据的索引结构。R-tree索引可以在地理信息系统（GIS）中使用，用于查询特定区域内的位置数据。
Full-text索引：一种用于对文本列进行全文搜索的索引结构。Full-text索引可以在文本中查找关键字和短语，并返回匹配的行。Full-text索引适合于对大量文本数据进行搜索和过滤的场景。
1.2 分析工具EXPLAIN
例子：
explain select * from person where age = 9
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ref | person_age_IDX | person_age_IDX | 5 | const | 1 | 100.0 |
| 字段 | 含义 |
|---|---|
| id | 执行select查询语句的序号，它是sql执行的顺序的标识，sql按照id从大到小执行，id相同的为一组，从上到下执行 |
| select_type | 查询的类型，也就是对应的是简单查询还是复杂查询 |
| table | 表名 |
| partitions | 当前查询匹配记录的分区。对于未分区的表，返回null |
| type | 连接类型，有如下几种取值，性能从好到坏排序：system>const>eq_ref>ref>index_merge>range>index>ALL |
| possible_keys | 展示当前查询可以使用哪些索引，这一列的数据是在优化过程的早期创建的，因此有些索引可能对于后续优化过程是没用的 |
| key | 表示MySQL实际选择的索引 |
| key_len | 索引使用的字节数。由于存储格式，当字段允许为NULL时，key_len比不允许为空时大1字节 |
| rows | MySQL估算会扫描的行数，数值越小越好 |
| ref | 表示将哪个字段或常量和key列所使用的字段进行比较 |
| filtered | 表示符合查询条件的数据百分比，最大100。用rows × filtered可获得和下一张表连接的行数。例如rows = 1000，filtered = 50%，则和下一张表连接的行数是500 |
| Extra | 展示有关本次查询的附加信息 |
1.2、优化器跟踪(Optimizer Trace)
这个优化器跟踪的目的是生成可被人类和程序读取的输出，以帮助理解MySQL优化器所采取的决策和操作。
开启操作语法：
SET SESSION OPTIMIZER_TRACE="enabled=on"; # enable tracing
<statement to trace>; # like SELECT, EXPLAIN SELECT, UPDATE, DELETE...
SELECT * FROM information_schema.OPTIMIZER_TRACE;
[ repeat last two steps at will ]
SET SESSION OPTIMIZER_TRACE="enabled=off"; # disable tracing
下面举个例子：
该例子用的数据下面person表中的，可以参考2、数据准备。
explain select * from person where age not in (9);
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ALL | person_age_IDX | 7 | 100.0 | Using where |
优化选择过程：
SET optimizer_trace="enabled=on";
select * from person where age not in (9);
SELECT * FROM information_schema.OPTIMIZER_TRACE;
SET optimizer_trace="enabled=off";
结果：
从分解过程中可以看到best_access_path中使用了scan，cost为2.4。
join_preparation：完成SQL的准备工作。在这个阶段，SQL语句会被格式化输出，通配符*会被具体字段代替，但不会进行等价改写动作。传入的SQL语句是select * from person where age not in (9)的结果。在完成了语句的补充、格式化后，准备阶段结束并进入下一阶段。
join_optimization：完成SQL语句的逻辑与物理优化的过程，这其中的优化步骤比较多。在展开具体内容之前，先解释下”select #”的问题。在输出中经常会看到有”select#:N”的字样，它表示当前跟踪的结构体是属于第几个SELECT。如果语句中使用多个SELECT语句拼接（如UNION）或者有嵌套子查询中有SELECT，会产生多个序号。
considered_execution_plans ：对比实际的不同路径的成本。如果是多表关联，且有存在执行顺序(如left/right join或straight_join来强制指定顺序)，则在plan_prefix部分会有前置条件；否则，就按照所有可能性评估。
{
"steps": [
{
"join_preparation": {
"select#": 1,
"steps": [
{
"expanded_query": "/* select#1 */ select `person`.`id` AS `id`,`person`.`age` AS `age`,`person`.`code` AS `code`,`person`.`name` AS `name`,`person`.`weight` AS `weight`,`person`.`height` AS `height`,`person`.`deleted` AS `deleted` from `person` where (`person`.`age` <> 9) limit 0,200"
}
]
}
},
{
"join_optimization": {
"select#": 1,
"steps": [
{
"condition_processing": {
"condition": "WHERE",
"original_condition": "(`person`.`age` <> 9)",
"steps": [
{
"transformation": "equality_propagation",
"resulting_condition": "(`person`.`age` <> 9)"
},
{
"transformation": "constant_propagation",
"resulting_condition": "(`person`.`age` <> 9)"
},
{
"transformation": "trivial_condition_removal",
"resulting_condition": "(`person`.`age` <> 9)"
}
]
}
},
{
"substitute_generated_columns": {
}
},
{
"table_dependencies": [
{
"table": "`person`",
"row_may_be_null": false,
"map_bit": 0,
"depends_on_map_bits": [
]
}
]
},
{
"ref_optimizer_key_uses": [
]
},
{
"rows_estimation": [
{
"table": "`person`",
"range_analysis": {
"table_scan": {
"rows": 7,
"cost": 4.5
},
"potential_range_indexes": [
{
"index": "PRIMARY",
"usable": false,
"cause": "not_applicable"
},
{
"index": "person_age_IDX",
"usable": true,
"key_parts": [
"age",
"id"
]
},
{
"index": "person_code_IDX",
"usable": false,
"cause": "not_applicable"
},
{
"index": "person_height_IDX",
"usable": false,
"cause": "not_applicable"
}
],
"setup_range_conditions": [
],
"group_index_range": {
"chosen": false,
"cause": "not_group_by_or_distinct"
},
"analyzing_range_alternatives": {
"range_scan_alternatives": [
{
"index": "person_age_IDX",
"ranges": [
"NULL < age < 9",
"9 < age"
],
"index_dives_for_eq_ranges": true,
"rowid_ordered": false,
"using_mrr": false,
"index_only": false,
"rows": 7,
"cost": 10.41,
"chosen": false,
"cause": "cost"
}
],
"analyzing_roworder_intersect": {
"usable": false,
"cause": "too_few_roworder_scans"
}
}
}
}
]
},
{
"considered_execution_plans": [
{
"plan_prefix": [
],
"table": "`person`",
"best_access_path": {
"considered_access_paths": [
{
"rows_to_scan": 7,
"access_type": "scan",
"resulting_rows": 7,
"cost": 2.4,
"chosen": true
}
]
},
"condition_filtering_pct": 100,
"rows_for_plan": 7,
"cost_for_plan": 2.4,
"chosen": true
}
]
},
{
"attaching_conditions_to_tables": {
"original_condition": "(`person`.`age` <> 9)",
"attached_conditions_computation": [
],
"attached_conditions_summary": [
{
"table": "`person`",
"attached": "(`person`.`age` <> 9)"
}
]
}
},
{
"refine_plan": [
{
"table": "`person`"
}
]
}
]
}
},
{
"join_execution": {
"select#": 1,
"steps": [
]
}
}
]
}
2、数据准备
| version() |
|---|
| 5.7.28-log |
创建数据库：
CREATE DATABASE `demo` /*!40100 DEFAULT CHARACTER SET latin1 */;
创建表：age和height为普通索引，code、name、weight为组合索引。
-- demo.person definition
CREATE TABLE `person` (
`id` bigint(20) NOT NULL AUTO_INCREMENT COMMENT '主键id',
`age` int(11) DEFAULT NULL COMMENT '年龄',
`code` varchar(100) DEFAULT NULL COMMENT '编码',
`name` varchar(50) DEFAULT NULL COMMENT '名字',
`weight` int(3) DEFAULT NULL COMMENT '体重',
`height` int(3) DEFAULT NULL COMMENT '身高',
`deleted` tinyint(1) NOT NULL DEFAULT '0' COMMENT '是否删除 0 1',
PRIMARY KEY (`id`),
KEY `person_age_IDX` (`age`) USING BTREE,
KEY `person_code_IDX` (`code`,`name`,`weight`) USING BTREE,
KEY `person_height_IDX` (`height`) USING BTREE
) ENGINE=InnoDB AUTO_INCREMENT=0 DEFAULT CHARSET=utf8mb4;
插入数据：
INSERT INTO demo.person (age, code, name, weight, height, deleted) VALUES(12, '4545345345342', '奥特曼', 15, 89, 0);
INSERT INTO demo.person (age, code, name, weight, height, deleted) VALUES(52, '552421231332', '卫斯理', 21, 45, 0);
INSERT INTO demo.person (age, code, name, weight, height, deleted) VALUES(87, '85124561321', '福尔摩斯', 50, 180, 0);
INSERT INTO demo.person (age, code, name, weight, height, deleted) VALUES(56, '86454212', '爱因斯坦', 56, 190, 0);
INSERT INTO demo.person (age, code, name, weight, height, deleted) VALUES(45, '86454212', '爱威立雅', 108, 172, 0);
INSERT INTO demo.person (age, code, name, weight, height, deleted) VALUES(10, '8754534', '爱尔兰', 78, 169, 0);
INSERT INTO demo.person (age, code, name, weight, height, deleted) VALUES(9, '867564233', '小兰', 98, 150, 0);
3、举例说明
1、普通索引
explain select * from person
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ALL | 7 | 100.0 |
索引有效=：
explain select * from person where age = 9
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ref | person_age_IDX | person_age_IDX | 5 | const | 1 | 100.0 |
发生值类型转换：'9'，依然有效。有些博文中说这个不行，万事没有绝对。
explain select * from person where age = '9'
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ref | person_age_IDX | person_age_IDX | 5 | const | 1 | 100.0 |
索引失效>：
explain select * from person where age > 9
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ALL | person_age_IDX | 7 | 85.71 | Using where |
索引有效in：
explain select * from person where age in(9)
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ref | person_age_IDX | person_age_IDX | 5 | const | 1 | 100.0 |
in索引不一定有效，如下面的情形，in的范围比较大时，索引失效。
explain select * from person where age in (9,10,28,56,41,50,92,20,12,51,54,2,45,2)
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ALL | person_age_IDX | 7 | 100.0 | Using where |
索引失效not in：
explain select * from person where age not in (9)
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ALL | person_age_IDX | 7 | 100.0 | Using where |
2、组合索引（联合摄影）
该组合索引是：code,name,weight
最左匹配原则：
MySQL 建立多列索引（联合索引）有最左匹配的原则，即最左优先：如果有一个 2 列的索引 (a, b)，则已经对 (a)、(a, b) 上建立了索引；如果有一个 3 列索引 (a, b, c)，则已经对 (a)、(a, b)、(a, b, c) 上建立了索引。
只有code 索引有效：
explain select * from person where code = '11222'
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ref | person_code_IDX | person_code_IDX | 403 | const | 1 | 100.0 |
name 索引失效：
explain select * from person where name = '小兰'
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ALL | 7 | 14.29 | Using where |
顺序为code,name时有效：
explain select * from person where code = '111' and name = '小兰'
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ref | person_code_IDX | person_code_IDX | 606 | const,const | 1 | 100.0 |
顺序为name,code时也有效，优化器会优化为：code,name。
explain select * from person where name = '111' and code = '小兰'
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ref | person_code_IDX | person_code_IDX | 606 | const,const | 1 | 100.0 |
优化器分析过程：
{
"steps": [
{
"join_preparation": {
"select#": 1,
"steps": [
{
"expanded_query": "/* select#1 */ select `person`.`id` AS `id`,`person`.`age` AS `age`,`person`.`code` AS `code`,`person`.`name` AS `name`,`person`.`weight` AS `weight`,`person`.`height` AS `height`,`person`.`deleted` AS `deleted` from `person` where ((`person`.`name` = '111') and (`person`.`code` = '小兰')) limit 0,200"
}
]
}
},
{
"join_optimization": {
"select#": 1,
"steps": [
{
"condition_processing": {
"condition": "WHERE",
"original_condition": "((`person`.`name` = '111') and (`person`.`code` = '小兰'))",
"steps": [
{
"transformation": "equality_propagation",
"resulting_condition": "((`person`.`name` = '111') and (`person`.`code` = '小兰'))"
},
{
"transformation": "constant_propagation",
"resulting_condition": "((`person`.`name` = '111') and (`person`.`code` = '小兰'))"
},
{
"transformation": "trivial_condition_removal",
"resulting_condition": "((`person`.`name` = '111') and (`person`.`code` = '小兰'))"
}
]
}
},
{
"substitute_generated_columns": {
}
},
{
"table_dependencies": [
{
"table": "`person`",
"row_may_be_null": false,
"map_bit": 0,
"depends_on_map_bits": [
]
}
]
},
{
"ref_optimizer_key_uses": [
{
"table": "`person`",
"field": "code",
"equals": "'小兰'",
"null_rejecting": false
},
{
"table": "`person`",
"field": "name",
"equals": "'111'",
"null_rejecting": false
}
]
},
{
"rows_estimation": [
{
"table": "`person`",
"range_analysis": {
"table_scan": {
"rows": 7,
"cost": 4.5
},
"potential_range_indexes": [
{
"index": "PRIMARY",
"usable": false,
"cause": "not_applicable"
},
{
"index": "person_age_IDX",
"usable": false,
"cause": "not_applicable"
},
{
"index": "person_code_IDX",
"usable": true,
"key_parts": [
"code",
"name",
"weight",
"id"
]
},
{
"index": "person_height_IDX",
"usable": false,
"cause": "not_applicable"
}
],
"setup_range_conditions": [
],
"group_index_range": {
"chosen": false,
"cause": "not_group_by_or_distinct"
},
"analyzing_range_alternatives": {
"range_scan_alternatives": [
{
"index": "person_code_IDX",
"ranges": [
"小兰 <= code <= 小兰 AND 111 <= name <= 111"
],
"index_dives_for_eq_ranges": true,
"rowid_ordered": false,
"using_mrr": false,
"index_only": false,
"rows": 1,
"cost": 2.21,
"chosen": true
}
],
"analyzing_roworder_intersect": {
"usable": false,
"cause": "too_few_roworder_scans"
}
},
"chosen_range_access_summary": {
"range_access_plan": {
"type": "range_scan",
"index": "person_code_IDX",
"rows": 1,
"ranges": [
"小兰 <= code <= 小兰 AND 111 <= name <= 111"
]
},
"rows_for_plan": 1,
"cost_for_plan": 2.21,
"chosen": true
}
}
}
]
},
{
"considered_execution_plans": [
{
"plan_prefix": [
],
"table": "`person`",
"best_access_path": {
"considered_access_paths": [
{
"access_type": "ref",
"index": "person_code_IDX",
"rows": 1,
"cost": 1.2,
"chosen": true
},
{
"access_type": "range",
"range_details": {
"used_index": "person_code_IDX"
},
"chosen": false,
"cause": "heuristic_index_cheaper"
}
]
},
"condition_filtering_pct": 100,
"rows_for_plan": 1,
"cost_for_plan": 1.2,
"chosen": true
}
]
},
{
"attaching_conditions_to_tables": {
"original_condition": "((`person`.`name` = '111') and (`person`.`code` = '小兰'))",
"attached_conditions_computation": [
],
"attached_conditions_summary": [
{
"table": "`person`",
"attached": null
}
]
}
},
{
"refine_plan": [
{
"table": "`person`"
}
]
}
]
}
},
{
"join_execution": {
"select#": 1,
"steps": [
]
}
}
]
}
code,name,weight顺序索引有效：
explain select * from person where code = '111' and name = '小兰' and weight = 52
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ref | person_code_IDX | person_code_IDX | 611 | const,const,const | 1 | 100.0 |
order by 索引有效：
explain select * from person where code = '111' and name = '小兰' and weight = 52 order by code,name,weight
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ref | person_code_IDX | person_code_IDX | 611 | const,const,const | 1 | 100.0 |
explain select * from person where code = '111' and name = '小兰' and weight = 52 order by name,weight
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ref | person_code_IDX | person_code_IDX | 611 | const,const,const | 1 | 100.0 |
explain select * from person where code = '111' and name = '小兰' and weight = 52 order by weight,name
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ref | person_code_IDX | person_code_IDX | 611 | const,const,const | 1 | 100.0 |
explain select * from person where code = '111' and name = '小兰' and weight = 52 order by weight,age
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ref | person_code_IDX | person_code_IDX | 611 | const,const,const | 1 | 100.0 | Using index condition; Using filesort |
code,weight,name顺序索引有效，优化器会优化为code,name,weight：
explain select * from person where code = '111' and weight = 52 and name = '小兰'
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ref | person_code_IDX | person_code_IDX | 611 | const,const,const | 1 | 100.0 |
code,weight索引有效，索引有效值为 code：
explain select * from person where code = '111' and weight = 52
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ref | person_code_IDX | person_code_IDX | 403 | const | 1 | 14.29 | Using index condition |
code,weight>,name 索引有效，索引有效值为 code：
explain select * from person where code = '111' and weight > 52 and name = '小兰'
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | range | person_code_IDX | person_code_IDX | 611 | 1 | 100.0 | Using index condition |
length(code)包含函数，索引失效，但是这个不一定，要看数据版本，数据量等：
explain select * from person where length(code)>10
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ALL | 7 | 100.0 | Using where |
索引有效：
explain select count(1) from person where code = '86454212'
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ref | person_code_IDX | person_code_IDX | 403 | const | 2 | 100.0 | Using index |
索引有效：
explain select sum(age) from person where code = '86454212'
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ref | person_code_IDX | person_code_IDX | 403 | const | 2 | 100.0 |
索引有效：
explain select * from person where code like '86454212'
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | range | person_code_IDX | person_code_IDX | 403 | 2 | 100.0 | Using index condition |
索引无效：
explain select * from person where code like '%86454212'
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ALL | 7 | 14.29 | Using where |
索引有效：
explain select * from person where code like '86454212%'
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | range | person_code_IDX | person_code_IDX | 403 | 2 | 100.0 | Using index condition |
or索引无效：
explain select * from person where code = '111' or name = '小兰'
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ALL | person_code_IDX | 7 | 26.53 | Using where |
组合索引和普通索引一起，索引无效：
explain select * from person where code = '111' or age = 9
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | ALL | person_age_IDX,person_code_IDX | 7 | 26.53 | Using where |
普通索引和普通索引一起，索引无效：
explain select * from person where age = 9 or height = 90
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SIMPLE | person | index_merge | person_age_IDX,person_height_IDX | person_age_IDX,person_height_IDX | 5,5 | 2 | 100.0 | Using union(person_age_IDX,person_height_IDX); Using where |
3、总结
- 最佳左匹配法则（重点）
- 计算、函数、类型转换(自动或手动)导致索引失效
- 范围条件右边的列索引失效
- 不等于(!= 或者<>)导致索引失效
- is null可以使用索引，is not null无法使用索引
- like以通配符%开头索引失效（重点）
- OR 前后只要存在非索引的列，都会导致索引失效