So you're basically creating an in-memory copy of every table and creating an in-memory index of every table by every column (an equivalency index only, not good for '>' or '<' or 'like' comparisions). This is bound to suck up huge amounts of memory on a database of any significant size. But interesting :-)