如果需要快速读取速度，最好的选择是使用闭包表。闭包表为每个祖先/后代对包含一行。在你的例子中，闭包表是这样的

ancestor | descendant | depth
0        | 0          | 0
0        | 19         | 1
0        | 20         | 2
0        | 21         | 3
0        | 22         | 4
19       | 19         | 0
19       | 20         | 1
19       | 21         | 3
19       | 22         | 4
20       | 20         | 0
20       | 21         | 1
20       | 22         | 2
21       | 21         | 0
21       | 22         | 1
22       | 22         | 0

一旦有了这个表，分层查询就变得非常简单和快速。获取类别20的所有子类:

SELECT cat.* FROM categories_closure AS cl
INNER JOIN categories AS cat ON cat.id = cl.descendant
WHERE cl.ancestor = 20 AND cl.depth > 0

当然，无论何时使用这样的非规格化数据都有一个很大的缺点。您需要在类别表旁边维护闭包表。最好的方法可能是使用触发器，但是正确跟踪闭包表的插入/更新/删除有点复杂。与任何事情一样，您需要查看您的需求，并决定哪种方法最适合您。

编辑:请参阅问题:在关系数据库中存储层次数据的选项是什么?更多选项。不同的情况有不同的最佳解决方案。

从博客管理分层数据在MySQL

表结构

+-------------+----------------------+--------+
| category_id | name                 | parent |
+-------------+----------------------+--------+
|           1 | ELECTRONICS          |   NULL |
|           2 | TELEVISIONS          |      1 |
|           3 | TUBE                 |      2 |
|           4 | LCD                  |      2 |
|           5 | PLASMA               |      2 |
|           6 | PORTABLE ELECTRONICS |      1 |
|           7 | MP3 PLAYERS          |      6 |
|           8 | FLASH                |      7 |
|           9 | CD PLAYERS           |      6 |
|          10 | 2 WAY RADIOS         |      6 |
+-------------+----------------------+--------+

查询:

SELECT t1.name AS lev1, t2.name as lev2, t3.name as lev3, t4.name as lev4
FROM category AS t1
LEFT JOIN category AS t2 ON t2.parent = t1.category_id
LEFT JOIN category AS t3 ON t3.parent = t2.category_id
LEFT JOIN category AS t4 ON t4.parent = t3.category_id
WHERE t1.name = 'ELECTRONICS';

输出

+-------------+----------------------+--------------+-------+
| lev1        | lev2                 | lev3         | lev4  |
+-------------+----------------------+--------------+-------+
| ELECTRONICS | TELEVISIONS          | TUBE         | NULL  |
| ELECTRONICS | TELEVISIONS          | LCD          | NULL  |
| ELECTRONICS | TELEVISIONS          | PLASMA       | NULL  |
| ELECTRONICS | PORTABLE ELECTRONICS | MP3 PLAYERS  | FLASH |
| ELECTRONICS | PORTABLE ELECTRONICS | CD PLAYERS   | NULL  |
| ELECTRONICS | PORTABLE ELECTRONICS | 2 WAY RADIOS | NULL  |
+-------------+----------------------+--------------+-------+

大多数用户都曾经在SQL数据库中处理过层次数据，毫无疑问，他们知道层次数据的管理不是关系数据库的目的。关系数据库的表不是分层的(像XML一样)，而只是一个平面列表。层次数据具有亲子关系，在关系数据库表中不能自然地表示这种关系。 阅读更多

更多细节请参考博客。

编辑:

select @pv:=category_id as category_id, name, parent from category
join
(select @pv:=19)tmp
where parent=@pv

输出:

category_id name    parent
19  category1   0
20  category2   19
21  category3   20
22  category4   21

参考:如何在Mysql中做递归SELECT查询?

使用BlueM/tree php类在mysql中创建一个自关系表的树。

Tree和Tree\Node是PHP类，用于处理使用父ID引用分层结构的数据。一个典型的例子是关系数据库中的一个表，其中每个记录的“父”字段引用另一个记录的主键。当然，Tree不能只使用来自数据库的数据，而是使用任何数据:您提供数据，Tree使用它，而不管数据来自何处以及如何处理。阅读更多

下面是一个使用BlueM/tree的例子:

<?php 
require '/path/to/vendor/autoload.php'; $db = new PDO(...); // Set up your database connection 
$stm = $db->query('SELECT id, parent, title FROM tablename ORDER BY title'); 
$records = $stm->fetchAll(PDO::FETCH_ASSOC); 
$tree = new BlueM\Tree($records); 
...

这里没有提到的是，为每个项添加持久路径列，尽管它与第二种备选方案有点相似，但对于大型层次结构查询和简单的(插入、更新、删除)项来说不同且成本较低。

一些像:

id | name        | path
19 | category1   | /19
20 | category2   | /19/20
21 | category3   | /19/20/21
22 | category4   | /19/20/21/22

例子:

-- get children of category3:
SELECT * FROM my_table WHERE path LIKE '/19/20/21%'
-- Reparent an item:
UPDATE my_table SET path = REPLACE(path, '/19/20', '/15/16') WHERE path LIKE '/19/20/%'

优化路径长度和ORDER BY路径使用base36编码代替实际数值路径id

 // base10 => base36
 '1' => '1',
 '10' => 'A',
 '100' => '2S',
 '1000' => 'RS',
 '10000' => '7PS',
 '100000' => '255S',
 '1000000' => 'LFLS',
 '1000000000' => 'GJDGXS',
 '1000000000000' => 'CRE66I9S'

https://en.wikipedia.org/wiki/Base36

还通过对编码的id使用固定长度和填充来抑制斜杠'/'分隔符

详细优化说明如下: https://bojanz.wordpress.com/2014/04/25/storing-hierarchical-data-materialized-path/

TODO

构建一个函数或过程，以分割检索一个项的祖先的路径

列出第一个递归的子元素的简单查询:

select @pv:=id as id, name, parent_id
from products
join (select @pv:=19)tmp
where parent_id=@pv

结果:

id  name        parent_id
20  category2   19
21  category3   20
22  category4   21
26  category24  22

．.． 左连接:

select
    @pv:=p1.id as id
  , p2.name as parent_name
  , p1.name name
  , p1.parent_id
from products p1
join (select @pv:=19)tmp
left join products p2 on p2.id=p1.parent_id -- optional join to get parent name
where p1.parent_id=@pv

@tincot列出所有孩子的解决方案:

select  id,
        name,
        parent_id 
from    (select * from products
         order by parent_id, id) products_sorted,
        (select @pv := '19') initialisation
where   find_in_set(parent_id, @pv) > 0
and     @pv := concat(@pv, ',', id)

用Sql Fiddle在线测试并查看所有结果。

http://sqlfiddle.com/ !9 / a318e3/4/0

试试这些:

表定义:

DROP TABLE IF EXISTS category;
CREATE TABLE category (
    id INT AUTO_INCREMENT PRIMARY KEY,
    name VARCHAR(20),
    parent_id INT,
    CONSTRAINT fk_category_parent FOREIGN KEY (parent_id)
    REFERENCES category (id)
) engine=innodb;

实验行:

INSERT INTO category VALUES
(19, 'category1', NULL),
(20, 'category2', 19),
(21, 'category3', 20),
(22, 'category4', 21),
(23, 'categoryA', 19),
(24, 'categoryB', 23),
(25, 'categoryC', 23),
(26, 'categoryD', 24);

存储过程:

DROP PROCEDURE IF EXISTS getpath;
DELIMITER $$
CREATE PROCEDURE getpath(IN cat_id INT, OUT path TEXT)
BEGIN
    DECLARE catname VARCHAR(20);
    DECLARE temppath TEXT;
    DECLARE tempparent INT;
    SET max_sp_recursion_depth = 255;
    SELECT name, parent_id FROM category WHERE id=cat_id INTO catname, tempparent;
    IF tempparent IS NULL
    THEN
        SET path = catname;
    ELSE
        CALL getpath(tempparent, temppath);
        SET path = CONCAT(temppath, '/', catname);
    END IF;
END$$
DELIMITER ;

存储过程的包装器函数:

DROP FUNCTION IF EXISTS getpath;
DELIMITER $$
CREATE FUNCTION getpath(cat_id INT) RETURNS TEXT DETERMINISTIC
BEGIN
    DECLARE res TEXT;
    CALL getpath(cat_id, res);
    RETURN res;
END$$
DELIMITER ;

选择的例子:

SELECT id, name, getpath(id) AS path FROM category;

输出:

+----+-----------+-----------------------------------------+
| id | name      | path                                    |
+----+-----------+-----------------------------------------+
| 19 | category1 | category1                               |
| 20 | category2 | category1/category2                     |
| 21 | category3 | category1/category2/category3           |
| 22 | category4 | category1/category2/category3/category4 |
| 23 | categoryA | category1/categoryA                     |
| 24 | categoryB | category1/categoryA/categoryB           |
| 25 | categoryC | category1/categoryA/categoryC           |
| 26 | categoryD | category1/categoryA/categoryB/categoryD |
+----+-----------+-----------------------------------------+

过滤指定路径的行:

SELECT id, name, getpath(id) AS path FROM category HAVING path LIKE 'category1/category2%';

输出:

+----+-----------+-----------------------------------------+
| id | name      | path                                    |
+----+-----------+-----------------------------------------+
| 20 | category2 | category1/category2                     |
| 21 | category3 | category1/category2/category3           |
| 22 | category4 | category1/category2/category3/category4 |
+----+-----------+-----------------------------------------+