ceteri-mapred

ceteri-mapred copied to clipboard

Getting Started on Hadoop

Paco Nathan [email protected]

Silicon Valley Cloud Computing Meetup

http://www.meetup.com/cloudcomputing/calendar/13911740/

Mountain View, 2010-07-19

GitHub src repo: http://github.com/ceteri/ceteri-mapred

Presentation slides available here in Keynote format or online at SlideShare: doc/enron.key http://www.slideshare.net/pacoid/getting-started-on-hadoop

See the "WordCount" example at: bin/run_wc.sh

See the "Enron Email Dataset" demo at: bin/run_enron.sh

R statistics demo: thresh.R, thresh.tsv

Gephi graph demo: graph.gephi

to run your own code on Elastic MapReduce

1. create a bucket in S3
2. copy the Python scripts into a "src" folder there
3. determine some subset of the email message input
cat msgs.tsv | head -1000 > input
4. copy "input" to your S3 "src" folder
5. follow examples in slide deck, based on params below

Hadoop job flow 1 on Elastic MapReduce

-input s3n://ceteri-mapred/enron/src/input -output s3n://ceteri-mapred/enron/src/output -mapper '"python map_parse.py http://ceteri-mapred.s3.amazonaws.com/ stopwords"' -reducer '"python red_idf.py 2500"' -cacheFile s3n://ceteri-mapred/enron/src/map_parse.py#map_parse.py -cacheFile s3n://ceteri-mapred/enron/src/red_idf.py#red_idf.py -cacheFile s3n://ceteri-mapred/enron/src/stopwords#stopwords

Hadoop job flow 2 on Elastic MapReduce

-input s3n://ceteri-mapred/enron/src/output -output s3n://ceteri-mapred/enron/src/filter -mapper '"python map_filter.py"' -reducer '"python red_filter.py 0.0633"' -cacheFile s3n://ceteri-mapred/enron/src/map_filter.py#map_filter.py -cacheFile s3n://ceteri-mapred/enron/src/red_filter.py#red_filter.py

after downloading the partition file named "filter" from S3, then

run the following command to build a lexicon:

cat filter/part-* | sort -k1 -k4 -nr > lexicon