Actually there would be two reasons - performance, and reliability. In
particular, a database may need to be certain that a particular piece
of data has been written to secondary store.
Every time this topic comes up I find myself wondering why people don't
see the obvious - it doesn't matter whether you need raw partitions on
Linux to achieve these goals. If they were necessary on any of the
earlier platforms on which the product was developed, then the code will
have been written to use them. If porting to buffered partitions is
not trivial, then it doesn't matter if they're good enough to do the
job - the work won't get done unless there's a compelling business case
for supporting the platform.
...............................................................................
Peter Desnoyers : Midnight Networks Inc. 200 Fifth Avenue Waltham MA 02154
pjd@midnight.com : Ph. 617/890-1001 Fax -0028 The Best in Network Software