避免在 MySQL 中使用 OR 对简单联接进行完整 table 扫描

Question

我有这个架构

create table table1
(
    id        int          auto_increment primary key,
    name      varchar(2)   null,
    position  int          null,
);

create index table1_position
    on table1 (position);


create table table_2
(
    id              int auto_increment primary key,
    table1_id       int          null,
    position        int          null,
    constraint table_2_ibfk_1
        foreign key (table1_id) references table1 (id)
);

create index ix_table_2_position
    on table_2 (position);

create index table1_id
    on table_2 (table1_id);

所以我在每个 table 的列 position 上添加了两个索引。现在我需要在 BOTH table 中寻找一个位置范围（通过加入 then 并应用 OR 查询）

所以我有这个查询

SELECT *
FROM table_1
    INNER JOIN table_2 ON table_1.id = table_2.table1_id
WHERE table_1.position BETWEEN 5000 AND 5500
   OR table_2.position BETWEEN 5000 AND 5500

但是 Explain 查询输出给我全部（完整 table 扫描）

id             1
select_type    SIMPLE
table          table_1
partitions
type           ALL
possible_keys  PRIMARY,table1_position
key
key_len
ref
rows           9929
filtered       100.0
Extra

如果我更改为 AND 如果给我预期的范围索引扫描

SELECT *
FROM table_1
    INNER JOIN table_2 ON table_1.id = table_2.table1_id
WHERE table_1.position BETWEEN 5000 AND 5500
   AND table_2.position BETWEEN 5000 AND 5500

id             1
select_type    SIMPLE
table          table_1
partitions
type           range
possible_keys  PRIMARY,pos_idx2
key            pos_idx2
key_len        5
ref
rows           1
filtered       100.0
Extra          Using index condition

但我在这里需要 OR 语句...我如何才能 mysql 对 OR 语句使用范围扫描索引？我可以在这里改进我的索引吗（我考虑过 position 和 table1_id 上的多值索引 - 外键，但它没有帮助并且它执行了完整的 table 扫描).

Answer 1

您可能出于性能原因想要避免 table 扫描。

所以尝试将 OR 换成 UNION 操作。

首先，使用子查询获取所需的 table1.id 值集，就像这样。

    SELECT id FROM table_1 WHERE position BETWEEN 5000 AND 5500
    UNION
    SELECT table1_id FROM table_2 WHERE position BETWEEN 5000 AND 5500

UNION 的第二部分通过选择 FK 列从 table_2 中检索您需要的 table_1.id 值。

接下来，使用该子查询从 table1 获取行。

SELECT * FROM table_1
 WHERE id IN (
    SELECT id FROM table_1 WHERE position BETWEEN 5000 AND 5500
    UNION
    SELECT table1_id FROM table_2 WHERE position BETWEEN 5000 AND 5500           
)

为了加快速度，请在 table_2 上添加此复合索引。

CREATE INDEX ix_table_2_position_table1id 
          ON table_2 (position, table1_id);

请注意，table2 上的两个单列索引都对此查询没有用。

Answer 2

SELECT *
FROM table_1
INNER JOIN table_2 ON table_1.id = table_2.table1_id
WHERE table_1.position BETWEEN 5000 AND 5500

UNION ALL

SELECT *
FROM table_1
INNER JOIN table_2 ON table_1.id = table_2.table1_id
WHERE table_1.position NOT BETWEEN 5000 AND 5500
  AND table_2.position BETWEEN 5000 AND 5500

也测试

SELECT *
FROM table_1
INNER JOIN table_2 ON table_1.id = table_2.table1_id
WHERE NOT (    table_1.position NOT BETWEEN 5000 AND 5500
           AND table_2.position NOT BETWEEN 5000 AND 5500 )

Answer 3

OR 的性能问题通常可以用 UNION 解决：

SELECT *
FROM table_1
INNER JOIN table_2 ON table_1.id = table_2.table1_id
WHERE table_1.position BETWEEN 5000 AND 5500
UNION
SELECT *
FROM table_1
INNER JOIN table_2 ON table_1.id = table_2.table1_id
WHERE table_2.position BETWEEN 5000 AND 5500;

避免在 MySQL 中使用 OR 对简单联接进行完整 table 扫描

Avoid full table scan on a simple join using OR in MySQL

mysql

sql

indexing

join

sql-optimization