I want to do a image processing program, but, if I had read the whole image into account, then it is so big a piece of data is passed a few times (I am now writing program is passed to the Map to pass a picture, the size of the Map is passed to Reduce to the size of the image, and Reduce the inside function to transmit two or three times), it is a waste of time, but also a waste of space, I think is passed to the file to be processed directly to the address Map, Map the address to Reduce again, finally in the Reduce of correlation function to open the file again? Excuse me, can open the Map and Reduce the HDFS file? What if you can open?

CodePudding user response:

HDFS and normal operation, just on the map, or reduce that executes