Example Storm Topologies

Learn to use Storm!

Table of Contents

  • Getting started

  • Using storm-starter with Maven

  • Using storm-starter with IntelliJ IDEA

Getting started


First, you need java and git installed and in your user’s PATH. Also, two of the examples in storm-starter

require Python and Ruby.

Next, make sure you have the storm-starter code available on your machine. Git/GitHub beginners may want to use the

following command to download the latest storm-starter code and change to the new directory that contains the downloaded


$ git clone git:// && cd storm/examples/storm-starter

storm-starter overview

storm-starter contains a variety of examples of using Storm. If this is your first time working with Storm, check out

these topologies first:

  1. ExclamationTopology: Basic topology written in all Java

  2. WordCountTopology: Basic topology that makes use of multilang by

    implementing one bolt in Python

  3. ReachTopology: Example of complex DRPC on top of Storm

After you have familiarized yourself with these topologies, take a look at the other topopologies in

src/jvm/storm/starter/ such as RollingTopWords

for more advanced implementations.

If you want to learn more about how Storm works, please head over to the

Storm project page.

Using storm-starter with Maven

Install Maven

Install Maven (preferably version 3.x) by following

the Maven installation instructions.

Build and install Storm jars locally

If you are using the latest development version of Storm, e.g. by having cloned the Storm git repository,

then you must first perform a local build of Storm itself. Otherwise you will run into Maven errors such as

“Could not resolve dependencies for project org.apache.storm:storm-starter:<storm-version>-SNAPSHOT“.

# Must be run from the top-level directory of the Storm code repository
$ mvn clean install -DskipTests=true

This command will build Storm locally and install its jar files to your user’s $HOME/.m2/repository/. When you run

the Maven command to build and run storm-starter (see below), Maven will then be able to find the corresponding version

of Storm in this local Maven repository at $HOME/.m2/repository.

Running topologies with Maven

Note: All following examples require that you run cd examples/storm-starter beforehand.

storm-starter topologies can be run with the maven-exec-plugin. For example, to

compile and run WordCountTopology in local mode, use the command:

$ mvn compile exec:java -Dstorm.topology=storm.starter.WordCountTopology

You can also run clojure topologies with Maven:

$ mvn compile exec:java -Dstorm.topology=storm.starter.clj.word_count

In Windows parameter should be quoted, like this:

$ mvn compile exec:java "-Dstorm.topology=storm.starter.clj.word_count"

Packaging storm-starter for use on a Storm cluster

You can package a jar suitable for submitting to a Storm cluster with the command:

$ mvn package

This will package your code and all the non-Storm dependencies into a single “uberjar” (or “fat jar”) at the path


Example filename of the uberjar:

>>> target/storm-starter-0.9.3-incubating-SNAPSHOT-jar-with-dependencies.jar

You can submit (run) a topology contained in this uberjar to Storm via the storm CLI tool:

# Example 1: Run the RollingTopWords in local mode (LocalCluster)
$ storm jar storm-starter-*-jar-with-dependencies.jar storm.starter.RollingTopWords

# Example 2: Run the RollingTopWords in remote/cluster mode,
# under the name "production-topology"
$ storm jar storm-starter-*-jar-with-dependencies.jar storm.starter.RollingTopWords production-topology remote

Submitting a topology in local vs. remote mode:

It depends on the actual code of a topology how you can or even must tell Storm whether to run the topology locally (in

an in-memory LocalCluster instance of Storm) or remotely (in a “real” Storm cluster). In the case of

RollingTopWords, for instance, this can be done by passing command line


Topologies other than RollingTopWords – such as ExclamationTopology

– may behave differently, e.g. by always submitting to a remote cluster (i.e. hardcoded in a way that you, as a user,

cannot change without modifying the topology code), or by requiring a customized configuration file that the topology

code will parse prior submitting the topology to Storm. Similarly, further options such as the name of the topology may

be user-configurable or be hardcoded into the topology code. So make sure you understand how the topology of your

choice is set up and configured!

Running unit tests

Use the following Maven command to run the unit tests that ship with storm-starter. Unfortunately lein test does not

yet run the included unit tests.

$ mvn test

Using storm-starter with IntelliJ IDEA

Importing storm-starter as a project in IDEA

The following instructions will import storm-starter as a new project in IntelliJ IDEA.

  • Open File > Import Project… and navigate to the storm-starter directory of your storm clone (e.g.


  • Select Import project from external model, select “Maven”, and click Next.

  • In the following screen, enable the checkbox Import Maven projects automatically. Leave all other values at their

    defaults. Click Next.

  • Click Next on the following screen about selecting Maven projects to import.

  • Select the JDK to be used by IDEA for storm-starter, then click Next.

    • At the time of this writing you should use JDK 6.

    • It is strongly recommended to use Sun/Oracle JDK 6 rather than OpenJDK 6.

  • You may now optionally change the name of the project in IDEA. The default name suggested by IDEA is “storm-starter”.

    Click Finish once you are done.

0 個評論