23 lines
629 B
Markdown
23 lines
629 B
Markdown
Style #30
|
|
==============================
|
|
|
|
Very similar to style #12, but with an additional twist
|
|
|
|
Constraints:
|
|
|
|
- Input data is divided in chunks, similar to what an inverse multiplexer does to input signals
|
|
|
|
- A map function applies a given worker function to each chunk of data, potentially in parallel
|
|
|
|
- The results of the many worker functions are reshuffled in a way
|
|
that allows for the reduce step to be also parallelized
|
|
|
|
- The reshuffled chunks of data are given as input to a second map
|
|
function that takes a reducible function as input
|
|
|
|
Possible names:
|
|
|
|
- Double inverse multiplexer
|
|
- Map-reduce
|
|
- Hadoop style
|