Realtime processing of data streams

• Stream: unbounded sequence of tuples.
• Topology: network of spouts and bolts.
  • Spout: source stream.
  • Bolt: stream processing.

Architecture

Nimbus node (master): code distribution + worker allocation

Zookeeper nodes: coordination

Supervisor nodes: start and stop workers

understanding the parallelism of a storm topology

Example 1: Basic uppercase

Example 2: Word Count

# zookeeper
brew install zookeeper
zkServer start

# storm
brew install storm
storm nimbus
storm supervisor
storm ui

Local env (OSX)

http://storm.apache.org/