You obviously didn't read my followup. To clarify, I said that the
space would be reused for future inserts or updates. if there is index
data that logically fits into the space freed in the b-tree the index
data will be placed there. However, the extent will not be placed on
the freelist until all of the data in the entire block has been
deleted. This effectively causes fragmentation (and potential growth)
of the index and would call for a possible rebuild.
Has anyone else heard the substance of the problem. The poor guy is
actually having to rebuild the database on a regular basis due to
non-sensical assumptions made by the business (and developers). He
proved that by creating an index, the problem was partially solved.
Possible further SQL tuning would get them to where they need to be. Is
this not a clasic case of the tail wagging the dog. The business needs
to be convinced that their approach is ridiculous and will prove
unmanageable in the long term. IBDOGAN's only hope is to illustrate
that his solution is the correct one. That won't make him an enemy if
he approaches them correctly. The one thing that you can never let
happen is to let non DBA's tell you how to manage a database.
Apparently, that has already happened.
Received on Fri Jun 24 2005 - 15:58:44 CDT
