On Mon, Feb 22, 2016 at 10:54 AM, Buck Calabro <kc2hiz@xxxxxxxxx> wrote:
What value do you see in a column by column conversion to DDL?
We've analyzed at our existing database and normalized it as much as
possible. Our SQL tables have column aliases so that existing
programs can access the fields as if they were still DDS. Tables that
used to have redundant data were restructured to be join views.
I've been filling the DBA role and have been creating indexes to
improve table access performance.
The issue came to a head (I wasn't aware he was even doing this)
because he complained that a field we added to a table wasn't in the
LF the index is implemented as. I had to drop & recreate the index in
order to get the field in the table he was trying to use.
Previously, when we added a field to a PF and recompiled all the LF's
that were based on the file, the new field was in the LF. Now we
don't have to recompile the indexes because the system takes care of
that for us.
My position is that, even though you CAN read an index to get to table
data ... you SHOULDN'T because that's not what indexes are intended to
be used for.
I'm looking for some IBM manual (or blog post) that makes that
statement. I've found lots of pages that state exactly that ... but
nothing (yet) from IBM.
Rob
midrange.com
