PostgreSQL Database Schema – How to Review and Optimize Design

You might need an intermediate table to store this relationship, as it's possible that a student has more than one parent/gaurdian (a mother and a father) and it's possible that a parent/gaurdian has more than on student.

student
-------
  id


gaurdian
--------
  id

student_gaurdian
----------------
  student_id
  gaurdian_id

The table student_gaurdian can store all the relations between students and gaurdians, and allows many-to-many relations.

RE: your new question

Having two tables with keys to each other looks a little weird to me, but if you really are having performance problems and this relieves them, then it's fine.

Assuming your surrogate keys are auto-incrementing integers marked as primary keys they will be the "rowid" for that table so you can read the last value added for a single insert using SELECT last_insert_rowid(). This operates like scope_identity in other databases. You would need to read each one after each row insert as each will replace record of the previous one.

sqlite doesn't seem to have an equivalent to the OUTPUT clause found in SQL Server and the RETURNING clause in Postgres & Oracle, so I don't know of a way to read multiple generated ID meaning that you'll have to do this one row at a time.