About MapReduce output directory little doubt

Newbie to heroes Neighborhoods:
1. output directory does not already exist , then the directory is not always MapReduce must delete this directory before ?
2. outputs can be specified directory can not be designated as a specific file name?
------ Solution ---------------------------------------- ----
1. If it is the same job , then you must remove the resulting output directory , which is a protective measure hadoop .
2. could not know the file name , you think you know a few PB of data in a meaningful filenames Well
------ Solution --------------- -----------------------------

Oh necessarily small
If you want 1T data row a sequence and then output
output file should 1T it
------ For reference only -------------------------- -------------

Thank you answer ah ! But the second point is still somewhat understand it, the output results do not quite small

------ For reference only ---------------------------------- -----
original so thank two of