All:
I am very excited about this product after glancing through the
documentation. The object-relational capabilities seem very
intriguing. I have one substantial concern, however. Am I correct in
that the creation of a database is limited to 1 file? I am used to the
Sybase and Oracle world where databases span several disks in order to
leverage disk drive to controller relationships. I would think without
this capability there is the potential for major disk thrashing in
something as simple as a query for which the SQL optimizer chooses to
use an index. Chances are, I assume, the index would be on one portion
of the drive whereas the data would be on another. The ability to have
the index on a separate drive than the data would seem to be
monumentally beneficial. I realize the intent for OLTP transaction
mixes is that the data would live in an in-memory cache, however, this
can't be guaranteed for very large databases.
Am I missing Something?