Cluster Primary Key in MySQL InnoDB

Question

We have a MySQL Database with a InnoDB table containing two integers fields let say A and B . Now suppose we create a clustered Primary Key using both indexes A and B. Everytime the pair A,B is populated we'll have a different index in our DB, so

• A=123 and B=456 will produce a unique index,
• A=123 and B=457 will produce another unique index,
• A=124 and B= 456 will produce another index and so on...

I was wondering if the situation A=123, B=456 will generate the same index (so is in collision) with A=456, B=123 . If the answer is NO (as I guess) how MySQL combine those fields in order to obtain a unique index? To generate the unique index, the sum A+B is not good for the reason above, the product A*B will waste a lot of space in variables furthermore will fail when just A or B is zero, so what's the algorithm used by MySQL (or any other DB) in order to achieve this result?

Answer 1

MySQL indexes preserve column order, and store the value for each column separately.

You can think of it being like a telephone book. The phone book is like an index over two columns: last name, then first name.

In the phone book, they don't add the names together. They store both, separated by a comma. So "Thomas Jay" is a different entry from "Jay Thomas." It's clear that they are separate fields.

It's the same in a multi-column index. The position of the columns matters. They are not added together or multiplied or anything else.

PS: Your use of "clustered" is a little bit confusing, because a clustered index already has a different meaning in databases. It means the table is stored in index order. I would suggest using "compound index" when you mean an index with more than one column.