Generally in real life tables called SRC have hundreds columns and millions of rows per months (part column). It must be prepared (transposed, same way as WIDE view does) to be processed using another procedure, which needs values from many months in one row. Typically we touch 6, sometimes 12 months. Sometimes there are more than one SRC table, so if there as eight SRC tables with 12 months WIDE view will join 96 parititions. If DB will get first part, join second, next db will join third to result, next woll join another, and again, again it will be veeery inefficient. Best way is to do sort and merge all tables i one pass.
Maybe view is not good option, what do you think?
↧