There is a difference between MySQL and Oracle here.
Oracle, to reduce parse/planner costs, hashes statements to see if it can
match an existing optimizer plan. This is optional and there are a few
flavors that range from a characher to characyter match through parse tree
matches through replacing
I'm about to do a whole bunch of testing here on various DA element
sizes, and datablock sizes and how the affect pg performance. It doesn't
appear possible to get > 4kb filesystem blocks under linux due to the
limitation of the pagesize. We're running AMD64 for these tests, but the
DA configur