erlang-mapreduce

Distributed Erlang implementation of letter-combination counting using map-reduce algorithm

Running

To run, you have several options (this is for timing purposes mostly) All runs will output two result files - results.txt and probabilities.txt. When no file name specified - assumes that the split up files (a-z) already exist from previous runs. This means you must first run mapreduce:main([],"354984si.ngl")., and only after that can you run the other styles without a filename specified.

1. Single node, no spawning with NO file-splitting (for benchmarking) run mapreduce:onlyOne("filename.txt").
2. Single node/multinode with NO file splitting (for checking different timing issues): run: mapreduce:main(). or mapreduce:main(['a@Computer','b@Computer',…]). This DOESN'T create onlya.txt through onlyz.txt but assumes they exist!
3. Single node/multinode with file name - meaning a file will now be split up. This is useful for the first time run: mapreduce:main([],"354984si.ngl"). or mapreduce:main(['a@Computer','b@Computer',...],"354984si.ngl"). This also creates the onlya.txt through onlyz.txt required.

Example Run

Example test runs:

 1> timer:tc(mapreduce,onlyOne,["354984si.ngl"]).
 {7379000,true}

 (a@Jengapad)1> timer:tc(mapreduce,main,[['a@Jengapad']]).
 {6238000,ok}

 (a@Jengapad)2> timer:tc(mapreduce,main,[['a@Jengapad','b@Jengapad']]).
 {5476000,ok}

 (a@Jengapad)3> timer:tc(mapreduce,main,[['a@Jengapad','b@Jengapad','c@Jengapad']]
 {4715000,ok}

 (a@Jengapad)4> timer:tc(mapreduce,main,[['a@Jengapad','b@Jengapad','c@Jengapad','d@Jengapad']]).
 {5434000,ok}

 (a@Jengapad)2> timer:tc(mapreduce,main,[['a@Jengapad','b@raspbmc','c@Jengapad','d@Jengapad']]).
 {16023000,ok}