While possible with older SQL's through your own code, distributed sharding and keeping multiple databases in sync I would think be useful at a DB level vs user code level.
You can certainly argue that shouldn't be part of the database software though.