An attempt do create real MapReduce with luigi

Create a map-reduce job to find the max number for the 1st field in a large csv file

• 1x mapper:
  • input: 0.large-file.txt
  • output: 1.part-00.txt … part-09.txt
• 10x transformer:
  • input: part-xx.txt
  • output: 2.transformed-xx.txt
• 10x reducer:
  • input: transformed-xx.txt
  • output: 3.solved-xx.txt
• 1x collector:
  • input: solved-xx.txt
  • output: 4.solution.txt

1. All transformers and groupers should run in parallel [resolved in 3.luigi-mapreduce ]
2. Transformers and groups should start working before mapper finishes [resolved with a hack in 4.luigi-mapreduce-smooth]