Sql-server – Why aren’t primary key / foreign key matches used for joins

When it comes to querying, indexing of a table should never be your first concern.

The queries you plan to use should dictate the indexes you need.

Based on the queries, some columns can be individually indexed. Other queries require compound indexes. The ORDER BY and GROUP BY clauses should provide immediate hints for indexes to make. Not using such hints may result in temp table sorting rather than using the indexes for data in the desired order needed.

Low cardinality of column values should eliminate the need for an index.

Even with these things taken into consideration, you may find that query may need some adjustment (a.k.a. refactoring) for performance gains.

When you reach the point of having the right indexes, not you have to worry about the size of those indexes. For a MyISAM table, this would mean that the .MYI file may grow significantly.

The size of the index file as well as the number of indexes should now be weighed against the performance of your queries, especially if the indexes provide the proper ordering of data and fastest retrieval.

Explain plans for queries may change over time depending on the number of rows, cardinality of columns, number of DELETEs and UPDATEs. Once a query's explain plan changes from what it looked like months ago, you should explore the need to add or remove indexes.

Reading your question literally, "if we delete 1 of 2 records pointing to Table2 from Table1 would also delete referenced record from Table2", you just need an ordinary ON DELETE CASCADE, as you've shown: Table2 references Table1, and deleting a record in Table1 causes all related records in Table2 to be deleted, even if those records are being pointed to by other records in Table1.

However, if your goal is instead to delete from Table2 when there are no more associated records in Table1, whereas there are normally two such records, you'll need something more elaborate.

I think better in concrete terms, so based on your pre-edit question, I'll call Table1 Guardians (because "parents" can be confusing for SQL-heads when they're actually the child half of a relationship) and Table2 Students (similarly, avoiding the word "children"). The goal is to ensure that any students that do not have any guardians get deleted.

If polygamy and divorce are unheard on, one option would be to have two NULLable fields in Students, FatherID and MotherID (or Parent1 and Parent2, to allow for step-parents, gay marriage, etc.). Set these to reference Guardians with ON DELETE SET NULL, and add an ON UPDATE trigger to Students to delete a record if both fields become NULL. If Bob's father is deleted but he still has a mother, he's OK, but if his mother is deleted the poor orphan gets wiped from the database. This assumes that a CASCADE UPDATE event triggers triggers, I don't know that for a fact.

A more robust solution would be to create a third table relating Guardians to Students. It would include fields GuardianID, StudentID, EffectiveDate (these three making up a candidate key), StopDate, and maybe fields to indicate the rights of the guardian; perhaps only one has legal custody, and the other should not be allowed to remove the student from school grounds. This would allow for death or divorce: just set the old record's StopDate. Insert a new record if the remaining guardian ever remarries. You could create an ON UPDATE trigger and an ON DELETE trigger on this third table, checking to see if there are now any students which lack active guardians.

However, before putting too much engineering into this, consider: do you really need to cascade? Keep the referential integrity, of course, but just enforce deletion logic in your business logic layer. For that matter, does it even make sense to delete students? If you're asked how many students took a specific class in 2011, but you've deleted some because they've subsequently dis-enrolled, you'll have an undercount.