利用Django框架中select_related和prefetch_related函数对数据库查询优化

时间:2022-04-26 04:53:50

实例的背景说明

假定一个个人信息系统,需要记录系统中各个人的故乡、居住地、以及到过的城市。数据库设计如下:

利用Django框架中select_related和prefetch_related函数对数据库查询优化

Models.py 内容如下:
 

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
from django.db import models
 
class Province(models.Model):
 name = models.CharField(max_length=10)
 def __unicode__(self):
  return self.name
 
class City(models.Model):
 name = models.CharField(max_length=5)
 province = models.ForeignKey(Province)
 def __unicode__(self):
  return self.name
 
class Person(models.Model):
 firstname = models.CharField(max_length=10)
 lastname = models.CharField(max_length=10)
 visitation = models.ManyToManyField(City, related_name = "visitor")
 hometown = models.ForeignKey(City, related_name = "birth")
 living  = models.ForeignKey(City, related_name = "citizen")
 def __unicode__(self):
  return self.firstname + self.lastname

注1:创建的app名为“QSOptimize”

注2:为了简化起见,`qsoptimize_province` 表中只有2条数据:湖北省和广东省,`qsoptimize_city`表中只有三条数据:武汉市、十堰市和广州市

如果我们想要获得所有家乡是湖北的人,最无脑的做法是先获得湖北省,再获得湖北的所有城市,最后获得故乡是这个城市的人。就像这样:
 

?
1
2
3
4
5
>>> hb = Province.objects.get(name__iexact=u"湖北省")
>>> people = []
>>> for city in hb.city_set.all():
... people.extend(city.birth.all())
...

显然这不是一个明智的选择,因为这样做会导致1+(湖北省城市数)次SQL查询。反正是个反例,导致的查询和获得掉结果就不列出来了。
prefetch_related() 或许是一个好的解决方法,让我们来看看。
 

?
1
2
3
4
5
>>> hb = Province.objects.prefetch_related("city_set__birth").objects.get(name__iexact=u"湖北省")
>>> people = []
>>> for city in hb.city_set.all():
... people.extend(city.birth.all())
...

因为是一个深度为2的prefetch,所以会导致3次SQL查询:
 

?
1
2
3
4
5
6
7
8
9
10
11
12
SELECT `QSOptimize_province`.`id`, `QSOptimize_province`.`name`
FROM `QSOptimize_province`
WHERE `QSOptimize_province`.`name` LIKE '湖北省' ;
 
SELECT `QSOptimize_city`.`id`, `QSOptimize_city`.`name`, `QSOptimize_city`.`province_id`
FROM `QSOptimize_city`
WHERE `QSOptimize_city`.`province_id` IN (1);
 
SELECT `QSOptimize_person`.`id`, `QSOptimize_person`.`firstname`, `QSOptimize_person`.`lastname`,
`QSOptimize_person`.`hometown_id`, `QSOptimize_person`.`living_id`
FROM `QSOptimize_person`
WHERE `QSOptimize_person`.`hometown_id` IN (1, 3);

嗯…看上去不错,但是3次查询么?倒过来查询可能会更简单?
 

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
>>> people = list(Person.objects.select_related("hometown__province").filter(hometown__province__name__iexact=u"湖北省"))
 
SELECT `QSOptimize_person`.`id`, `QSOptimize_person`.`firstname`, `QSOptimize_person`.`lastname`,
`QSOptimize_person`.`hometown_id`, `QSOptimize_person`.`living_id`, `QSOptimize_city`.`id`,
`QSOptimize_city`.`name`, `QSOptimize_city`.`province_id`, `QSOptimize_province`.`id`, `QSOptimize_province`.`name`
FROM `QSOptimize_person`
INNER JOIN `QSOptimize_city` ON (`QSOptimize_person`.`hometown_id` = `QSOptimize_city`.`id`)
INNER JOIN `QSOptimize_province` ON (`QSOptimize_city`.`province_id` = `QSOptimize_province`.`id`)
WHERE `QSOptimize_province`.`name` LIKE '湖北省';
 
+----+-----------+----------+-------------+-----------+----+--------+-------------+----+--------+
| id | firstname | lastname | hometown_id | living_id | id | name | province_id | id | name |
+----+-----------+----------+-------------+-----------+----+--------+-------------+----+--------+
| 1 | 张  | 三  |   3 |   1 | 3 | 十堰市 |   1 | 1 | 湖北省 |
| 2 | 李  | 四  |   1 |   3 | 1 | 武汉市 |   1 | 1 | 湖北省 |
| 3 | 王  | 麻子  |   3 |   2 | 3 | 十堰市 |   1 | 1 | 湖北省 |
+----+-----------+----------+-------------+-----------+----+--------+-------------+----+--------+
3 rows in set (0.00 sec)

完全没问题。不仅SQL查询的数量减少了,python程序上也精简了。
select_related()的效率要高于prefetch_related()。因此,最好在能用select_related()的地方尽量使用它,也就是说,对于ForeignKey字段,避免使用prefetch_related()。
联用
对于同一个QuerySet,你可以同时使用这两个函数。
在我们一直使用的例子上加一个model:Order (订单)
 

?
1
2
3
4
5
6
class Order(models.Model):
 customer = models.ForeignKey(Person)
 orderinfo = models.CharField(max_length=50)
 time  = models.DateTimeField(auto_now_add = True)
 def __unicode__(self):
  return self.orderinfo

如果我们拿到了一个订单的id 我们要知道这个订单的客户去过的省份。因为有ManyToManyField显然必须要用prefetch_related()。如果只用prefetch_related()会怎样呢?
 

?
1
2
3
4
>>> plist = Order.objects.prefetch_related('customer__visitation__province').get(id=1)
>>> for city in plist.customer.visitation.all():
... print city.province.name
...

显然,关系到了4个表:Order、Person、City、Province,根据prefetch_related()的特性就得有4次SQL查询
 

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
SELECT `QSOptimize_order`.`id`, `QSOptimize_order`.`customer_id`, `QSOptimize_order`.`orderinfo`, `QSOptimize_order`.`time`
FROM `QSOptimize_order`
WHERE `QSOptimize_order`.`id` = 1 ;
 
SELECT `QSOptimize_person`.`id`, `QSOptimize_person`.`firstname`, `QSOptimize_person`.`lastname`, `QSOptimize_person`.`hometown_id`, `QSOptimize_person`.`living_id`
FROM `QSOptimize_person`
WHERE `QSOptimize_person`.`id` IN (1);
 
SELECT (`QSOptimize_person_visitation`.`person_id`) AS `_prefetch_related_val`, `QSOptimize_city`.`id`,
`QSOptimize_city`.`name`, `QSOptimize_city`.`province_id`
FROM `QSOptimize_city`
INNER JOIN `QSOptimize_person_visitation` ON (`QSOptimize_city`.`id` = `QSOptimize_person_visitation`.`city_id`)
WHERE `QSOptimize_person_visitation`.`person_id` IN (1);
 
SELECT `QSOptimize_province`.`id`, `QSOptimize_province`.`name`
FROM `QSOptimize_province`
WHERE `QSOptimize_province`.`id` IN (1, 2);
?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
+----+-------------+---------------+---------------------+
| id | customer_id | orderinfo  | time    |
+----+-------------+---------------+---------------------+
| 1 |   1 | Info of Order | 2014-08-10 17:05:48 |
+----+-------------+---------------+---------------------+
1 row in set (0.00 sec)
 
+----+-----------+----------+-------------+-----------+
| id | firstname | lastname | hometown_id | living_id |
+----+-----------+----------+-------------+-----------+
| 1 | 张  | 三  |   3 |   1 |
+----+-----------+----------+-------------+-----------+
1 row in set (0.00 sec)
 
+-----------------------+----+--------+-------------+
| _prefetch_related_val | id | name | province_id |
+-----------------------+----+--------+-------------+
|      1 | 1 | 武汉市 |   1 |
|      1 | 2 | 广州市 |   2 |
|      1 | 3 | 十堰市 |   1 |
+-----------------------+----+--------+-------------+
3 rows in set (0.00 sec)
 
+----+--------+
| id | name |
+----+--------+
| 1 | 湖北省 |
| 2 | 广东省 |
+----+--------+
2 rows in set (0.00 sec)

更好的办法是先调用一次select_related()再调用prefetch_related(),最后再select_related()后面的表
 

?
1
2
3
4
>>> plist = Order.objects.select_related('customer').prefetch_related('customer__visitation__province').get(id=1)
>>> for city in plist.customer.visitation.all():
... print city.province.name
...

这样只会有3次SQL查询,Django会先做select_related,之后prefetch_related的时候会利用之前缓存的数据,从而避免了1次额外的SQL查询:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
SELECT `QSOptimize_order`.`id`, `QSOptimize_order`.`customer_id`, `QSOptimize_order`.`orderinfo`,
`QSOptimize_order`.`time`, `QSOptimize_person`.`id`, `QSOptimize_person`.`firstname`,
`QSOptimize_person`.`lastname`, `QSOptimize_person`.`hometown_id`, `QSOptimize_person`.`living_id`
FROM `QSOptimize_order`
INNER JOIN `QSOptimize_person` ON (`QSOptimize_order`.`customer_id` = `QSOptimize_person`.`id`)
WHERE `QSOptimize_order`.`id` = 1 ;
 
SELECT (`QSOptimize_person_visitation`.`person_id`) AS `_prefetch_related_val`, `QSOptimize_city`.`id`,
`QSOptimize_city`.`name`, `QSOptimize_city`.`province_id`
FROM `QSOptimize_city`
INNER JOIN `QSOptimize_person_visitation` ON (`QSOptimize_city`.`id` = `QSOptimize_person_visitation`.`city_id`)
WHERE `QSOptimize_person_visitation`.`person_id` IN (1);
 
SELECT `QSOptimize_province`.`id`, `QSOptimize_province`.`name`
FROM `QSOptimize_province`
WHERE `QSOptimize_province`.`id` IN (1, 2);
 
+----+-------------+---------------+---------------------+----+-----------+----------+-------------+-----------+
| id | customer_id | orderinfo  | time    | id | firstname | lastname | hometown_id | living_id |
+----+-------------+---------------+---------------------+----+-----------+----------+-------------+-----------+
| 1 |   1 | Info of Order | 2014-08-10 17:05:48 | 1 | 张  | 三  |   3 |   1 |
+----+-------------+---------------+---------------------+----+-----------+----------+-------------+-----------+
1 row in set (0.00 sec)
 
+-----------------------+----+--------+-------------+
| _prefetch_related_val | id | name | province_id |
+-----------------------+----+--------+-------------+
|      1 | 1 | 武汉市 |   1 |
|      1 | 2 | 广州市 |   2 |
|      1 | 3 | 十堰市 |   1 |
+-----------------------+----+--------+-------------+
3 rows in set (0.00 sec)
 
+----+--------+
| id | name |
+----+--------+
| 1 | 湖北省 |
| 2 | 广东省 |
+----+--------+
2 rows in set (0.00 sec)

值得注意的是,可以在调用prefetch_related之前调用select_related,并且Django会按照你想的去做:先select_related,然后利用缓存到的数据prefetch_related。然而一旦prefetch_related已经调用,select_related将不起作用。

 小结

  1.     因为select_related()总是在单次SQL查询中解决问题,而prefetch_related()会对每个相关表进行SQL查询,因此select_related()的效率通常比后者高。
  2.     鉴于第一条,尽可能的用select_related()解决问题。只有在select_related()不能解决问题的时候再去想prefetch_related()。
  3.     你可以在一个QuerySet中同时使用select_related()和prefetch_related(),从而减少SQL查询的次数。
  4.     只有prefetch_related()之前的select_related()是有效的,之后的将会被无视掉。