Using a global sequence means 8 bytes less storage per row (including index entries and all foreign keys which refer to the PK), and more importantly typically

Using a global sequence means 8 bytes less storage per row (including index entries and all foreign keys which refer to the PK), and more importantly typically mostly sequential inserts into B-tree indexes which means less bloat and faster insert and access. Disadvantages are data leakage and, theoretically, minor contention on accessing the sequence.