7

I am starting to design a typical product based MySQL database but I keep running in circles on how to design the tables.

I have many types of products

  • Jeans
  • Tshirts
  • Dresses

Which have many of the same attributes

  • Size
  • Price
  • Colour

But some of the products have specific attributes, like length or logo or package.

Do I make a separate table for every product even though many of the columns are the same, or do I make a general product table and then have some sub-tables for specific products?

  • 1
    How many 'specific attributes' do you have? Is it an entirely known list? or are you going to get odd ball ones in there from time to time as new products are added? – user40980 Apr 13 '15 at 21:06
  • I would imagine sizing information is different for jeans, t-shirts and dresses too... – h.j.k. Apr 14 '15 at 1:23
  • The reason I ask this is that its one thing to have a bunch of specific clothing related details if you are, say, a clothing store. On the other hand, if you are a general retailer, that level of detail of clothing fields isn't going to be applicable to the toy department, or home improvement, or automotive, or ... And thus you will want a more general solution across the board. But it would help to understand the nature of the product areas and range of items to be added. – user40980 Apr 14 '15 at 3:24
8

Do not make a table for every product.

This problem has been solved many ways. Try this:

Make a product (or products) table, put your common product attributes in the product table, then make an attribute table and a productattribute table, something like:

attribute
---------
attributeId
attributeName
attributeDescription


productattribute
----------------
productId
attributeId
attributeValue

Then you can assign arbitrary attributes to your products.

You can take the schema a little further if you like, using:

attribute
---------
attributeId
attributeName
attributeDescription
attributeType // [float, int, varchar, etc.]


productAttribute
----------------
productId
attributeId
attributeValueFloat
attributeValueInt
attributeValueVarChar

This technique complicates your schema a bit, so be aware of what you are getting into if you decide to use the extra columns to store values by data type.

An alternative that might just suit your needs a little better would be to use a NoSQL Store like MongoDB for your products. Then you can store the attributes you need for each type of product without resorting to the complexity of a relational database schema.

1

One approach would be to design your database similar to how you would define a class hierarchy, where you define base table(s) (classes) that provide the common attributes, and then you add additional tables that provide specific attributes to extend the table to the specific products.

Define a base product table

create table Products_base as
    id
    item
    brand
    model
    description
    size
    color
    price #more about this later

Define additional tables for products that differ substantially

create table pants as
    base #foreign key to base product
    waist
    length
    material
    style

create table dress as
    base #foreign key to base product table
    waist
    hemlength
    bust
    material
    belt

Another approach would be to define the base product table, and then define an attributes table, and provide the attributes for each additional product,

create table Products_base as
    id autoincrement
    item
    brand
    model
    description
    size
    color
    price #more about this later

create table attributes as
    id autoincrement
    name
    description

create table product_+attributes as
    product_id #product.id
    attribute_id #attributes.id
    value

Using pants as an example, you would want to add attributes to pants (add to the attributes table)

"length", "length of item"
"inseam", "length of inseam"
"style", "style of article" #example, "jeans"
#etc

And then you would add values for the attributes

product.id("pants"), attribute.id("inseam"), "32in"
product.id("pants"), attribute.id("waist"), "34in"
product.id("pants"), attribute.id("style"), "jeans"
product.id("dress"), attribute.id("style"), "sundress"
#etc
  • Creating an extra table for every kind of products seems like a no-go. – Tulains Córdova Apr 14 '15 at 19:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.