Question

I am creating an event to user lookup table using the Innodb search engine.

Table Event_Users

event_id
user_id

Index (PRIMARY) -> event_id, user_id
Index (secondary) -> user_id, event_id ??

The application will search user to event and event to user. How do I define the second index if the primary key is clustered? Should it simply be user_id and then MYSQL will figure out that event_id is already present in the PK or should I be redundant and include user_id, event_id in the secondary index?

In addition, is this a good use of a clustered index? Any guidance and help with Innodb clustered indexes would be greatly appreciated. Thanks!

Answer 1

MySQL should be able to use the fields in the PK as a covering index without making a composite secondary index.

Therefore:

Table (Engine = INNODB)

column_a  (int)
column_b  (int)

Primary Key -> column_a, column_b

If you would like to search by column_b simply add index: column_b.

Since the table using the Innodb engine, the PK will exist in the column_b index. Mysql will be able to use the index (using index). No need to create index column_b, column_a as this would be redundant. – Remember, inndob clusters by PK and all indexes reference the PK.

A surrogate key is a poor choice for this design because extra work would have to be made to ensure column_a, column_b does not already exist.