How Linux 3.6 Nearly Broke PostgreSQL 14 years ago
Lack of support for row level locking was a problem. I recall having to work around that by padding records with extra columns to ensure only one row would fit on a page. But I'd say more generally the bigger issue was in their overall approach to concurrency. Oracle's optimistic currency control mechanism (where readers don't wait for writers) worked better in practice than Sybase's early lock-based concurrency control (where they did).
I recall Philip Greenspun dedicated a substantial portion of his late 90's database-backed website book to that topic.