Should I use hash or btree for a foreign key index in postgresql 9.3?

Question

Which index will perform better for foreign keys of type integer in postgresql 9.3?

I would assume a hash index, because foreign key comparisson are always made with =

Or does a btree compare as fast as a hash when used for JOINS on foreign keys?

Because in postgresql primary keys use btree's that would suggest they are also better for foreign keys.

Answer 1

From the manual On PostgreSQL 9.3:

Caution Hash index operations are not presently WAL-logged, so hash indexes might need to be rebuilt with REINDEX after a database crash if there were unwritten changes. Also, changes to hash indexes are not replicated over streaming or file-based replication after the initial base backup, so they give wrong answers to queries that subsequently use them. For these reasons, hash index use is presently discouraged.

There is also no proof that an hash index has any performance benefits over a btree.

Answer 2

It depends.

If you're not doing any queries using the foreign key, then you don't need any index. The referential integrity is enforced using the index of the referenced primary key.

The question of which index type to use (if any) is thus the same for any column(s). If your queries would benefit from an index for a = comparison, and you're using PostgreSQL 10 or newer, then a HASH index is a reasonable choice. If the same column is involved in any ordering operations (ORDER BY, <, >=, etc.) then you may as well use a BTREE.

If you are concerned about the relative performance, then you'll need to test them yourself, using your own data distribution and query load. A tree index may still perform better than a hash, due to data access locality (sequential vs. random).