i was just doing some reading and come upon this in the manuals :
"Updating statistics for leading columns in indexes requires only a leaf-level scan of the index, and does not require a sort, so updating statistics for these columns does not affect concurrent performance very much."
so if we have a million page table, are we to take it that 10 worker processes each working on 1/10th of the leaf pages of the table to compile distribution statistics of the leading column, will not be significantly faster than 1 process doing the whole thing?
and any particular reason that parallelism seem to be linked only to sorting? surely there are other kinds of processing that too can benefit from parallelism?