Hdfs and map reduce word count?

[#1]

can anyone explain pr direct my to material explaining how map reduce word count works? I do not understand how it can work!

If a file is split into blocks and distributed over multiple nodes how can the word count program work? The file/text can be split in the middle of a word fx "be" in one block and "tween" in another block. How can the map reduce job count between as word if it is split over multiple blocks and nodes??

[#2 · Accepted solution]

The data format will ensure records are intact before being sent to the
mapper function. I believe this is done by sending partial records to the
machine they will be mapped on (so the overwhelming majority of the data is
processed in place, but half a line per block or so may still be
exchanged). Tom White's book The Definitive Guide To Hadoop does a good job
of covering details like this.

[#3]

Ok very cool. Thank you for the reference.