when should i create a separate table for fields? - mysql

Question

Let's say i have this SQL table :
CREATE TABLE `article` (
`article_id` int(11) NOT NULL AUTO_INCREMENT,
`name` varchar(255) NOT NULL,
`family` varchar(255) NOT NULL,
`sub_family` varchar(255) CHARACTER SET utf8 COLLATE utf8_general_ci NOT NULL,
`price` float NOT NULL,
PRIMARY KEY (`article_id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8
Here every family of articles has one or multiple sub_family
should i create another table that has all possible combinations of family and sub_family and refer to it using a foreing key ?
Data example :

Answer 1

If you want to fix redundancy completely, please try the following table:
create table `family` (
`family_id` int(11) NOT NULL AUTO_INCREMENT,
`family` varchar(255) NOT NULL,
`parent_id` int(11) NOT NULL,
PRIMARY KEY (`family_id`),
index idx_family_sub (`family`, `parent_id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;
CREATE TABLE `article` (
`article_id` int(11) NOT NULL AUTO_INCREMENT,
`name` varchar(255) NOT NULL,
`family_id` int(11) not null,
`price` float NOT NULL,
PRIMARY KEY (`article_id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;
The data in there 2 table may look like:
mysql> select * from family order by family_id;
+-----------+-------------+-----------+
| family_id | family | parent_id |
+-----------+-------------+-----------+
| 1 | Main Course | 0 |
| 2 | Beverage | 0 |
| 3 | Pasta | 1 |
| 4 | Hot Drinks | 2 |
| 5 | Cold Drinks | 2 |
+-----------+-------------+-----------+
5 rows in set (0.00 sec)
mysql> select * from article;
+------------+-----------+-----------+-------+
| article_id | name | family_id | price |
+------------+-----------+-----------+-------+
| 1 | Spaghetti | 3 | 600 |
| 2 | Penne | 3 | 500 |
| 3 | Coffee | 4 | 100 |
| 4 | Penne | 5 | 150 |
+------------+-----------+-----------+-------+
4 rows in set (0.00 sec)
The query to get all data like data example is :
mysql> select
-> a.article_id, a.name, f2.family as family, f.family as sub_family, a.price
-> from `article2` a
-> join family2 f on a.family_id = f.family_id
-> join family2 as f2 on f.parent_id = f2.family_id;
+------------+-----------+-------------+-------------+-------+
| article_id | name | family | sub_family | price |
+------------+-----------+-------------+-------------+-------+
| 1 | Spaghetti | Main Course | Pasta | 600 |
| 2 | Penne | Main Course | Pasta | 500 |
| 3 | Coffee | Beverage | Hot Drinks | 100 |
| 4 | Penne | Beverage | Cold Drinks | 150 |
+------------+-----------+-------------+-------------+-------+
4 rows in set (0.00 sec)
From my perspective, this table is used to save food menu data, and it should be very small. I think it's OK to create a dedicated table to save family and sub_family.

Answer 2

I read your question and comments as well. I think you can create two different tables that'll for sure reduce the data redundancy and complexity.
You can create Family as a parent table:
In Family table you can declare a unique key(which you can use further for foreign key referencing) for each Family and sub-Family pair for example:
Family:
ID | Family_name | Subfamily
1 Beverage Hot Drinks
2 Beverage cold Drinks
3 main course Pasta
And you can create another table for Articles which will be connected with parent via foreign key reference ID.
Article:
ID | Family | Name | Price
1 2 Milk Shake 100.00
2 1 Coffee 200.00
You can try something like this. I hope you find this easy to understand.