Tweitgeist v1.2.0

Tweitgeist analyses the Twitter Spitzer hose and compute in realtime the top trending hashtags using RedStorm/Storm. What makes this interesting other than being a cool Storm example, is the fact that this architecture will work at full Twitter Firehose scale without much modifications.

• See the slideshare presentation about Twitter Big Data and Tweitgeist.
• See the live demo on http://tweitgeist.colinsurprenant.com/

There are three components:

• The Twitter Spitzer stream reader which pushes messages in a Redis queue
• The Redstorm analyser which read the Twitter stream queue, computes the trending hashtags and output the top N list every 5 seconds in a Redis queue
• The viewer UI for the visualization

Dependencies

This has been tested on OSX 10.6+, Linux 11.10 & 12.04 using JRuby 1.6.x for the RedStorm topology and Ruby 1.9.x for the Twitter Spitzer hose reader.

Installation

• Redis is required
• RVM is highly recommended as you will need to work with both Ruby/JRuby and different gemsets.

Redstorm backend

• requires JRuby 1.6.x

• set JRuby in 1.9 mode by default

  export JRUBY_OPTS=--1.9

• install the RedStorm gem using bundler with the supplied Gemfile

  $ bundle install

• run RedStorm installation

  $ bundle exec redstorm install

• package the topology required gems

  $ bundle exec redstorm bundle topology

• if you plan on running the topology on a cluster, package the topology jar

  bundle exec redstorm jar lib/tweitgeist/

Twitter Spitzer stream reader

• requires Ruby 1.9.x

• install required gems using bundler with the supplied Gemfile

  $ bundle install

Viewer

• requires Node.js

  $ sudo apt-get install nodejs

• requires npm

  $ sudo apt-get install npm

• install CoffeeScript if you want to modify the Node.js server

  $ npm install -g coffee-script

• install other dependencies

  $ cd lib/viewer
  $ npm install .

Usage overview

Redstorm backend

• requires JRuby 1.6.x

• set JRuby in 1.9 mode by default

  export JRUBY_OPTS=--1.9

RedStorm backend in local mode.

$ bundle exec redstorm local lib/tweitgeist/storm/tweitgeist_topology.rb

RedStorm backend in remote cluster mode.

• add your cluster info to ~/.storm/storm.yaml see setting up a Storm development environment

• make sure your locally installed storm distribution bin/ directory is in your $PATH

$  bundle exec redstorm cluster lib/tweitgeist/storm/tweitgeist_topology.rb

Twitter Spitzer stream reader

• requires Ruby 1.9.x

• edit config/twitter_reader.rb to add your credentials

$ ruby lib/tweitgeist/twitter/twitter_reader.rb

Viewer

$ coffee server.coffee --port 8080 --host 127.0.0.1 --redis-port 6379 --redis-host 127.0.0.1

or (with simulated data in case of no redis)

$ coffee server.coffee --port 8080 --host 127.0.0.1 --mock

Author

Colin Surprenant, @colinsurprenant, https://github.com/colinsurprenant, [email protected]

Contributors

Francois Lafortune, @quickredfox, https://github.com/quickredfox, [email protected]

Nicholas Brochu, @nbrochu, https://github.com/nbrochu, [email protected]

License

Tweitgeist is distributed under the Apache License, Version 2.0.