Architecture Overview

...

Overview

Kettle transformations are used to manipulate data and function as the map, combine, and reduce phases of a MapReduce application. The Kettle engine is pushed down to each task node and is executed for each task. The implementation that supports the data type conversion from Hadoop data types to Kettle data types, the passing of tuples between input/output formats to the Kettle engine, and all associated configuration for the MapReduce job is collectively called Pentaho MapReduce.

Type Mapping

In order to pass data between Hadoop and Kettle we must convert between Hadoop IO data types. Here's the type mapping for the built in Kettle types:

...

Defining your own Type Converter

TODO
See The Type Converter system is pluggable to support additional data types as required by custom Input/Output formats. The Type Converter SPI is a simple interface to implement: org.pentaho.hadoop.mapreduce.converter.spi.ITypeConverter for more info. We use the Service Locator pattern; specifically Java's ServiceLoader, to resolve available converters at runtime. Providing your own is as easy as implementing ITypeConverter and providing a META-INF/services/org.pentaho.hadoop.mapreduce.converter.spi.ITypeConverter file with your implementation listed, both packaged into a jar placed in the plugins/pentaho-big-data-plugin/lib directory. You can find the default implementations defined here.

Distributed Cache

Pentaho MapReduce relies on Hadoop's Distributed Cache to distribute the Kettle environment, configuration, and plugins across the cluster. By leveraging the Distributed Cache network traffic is reduced up for subsequent executions as the Kettle environment is automatically configured on each node. This also allows you to use multiple version of Kettle against a single cluster.

...

The default Kettle environment installation path within HDFS is /opt/pentaho/mapreduce/ pmr.kettle.dfs.install.dir/$id, where $id is generally the version of Kettle the environment contains a uniquely identifying string but can easily be a custom build that is tailored for a specific set of jobs.

The Kettle environment is staged to HDFS at pmr.kettle.dfs.install.dir/pmr.kettle.installation.id as follows:

1. The contents of plugins/pentaho-big-data-plugin/pentaho-mapreduce-libraries.zip are extracted into HDFS at hdfs://{pmr.kettle.dfs.install.dir}/{pmr.kettle.installation.id}
2. The Big Data Plugin contents are copied into pmr.kettle.installation.id/plugins/
   1. Only the active Hadoop configuration is copied, and specifically:
      1. The active Hadoop configuration's client-only libraries are not copied (config/lib/client)
      2. The active Hadoop configuration's "pmr" specific libraries are copied into the main hdfs://{pmr.kettle.dfs.install.dir}/{pmr.kettle.installation.id}/lib/ of the installation. This allows the Hadoop configuration to provide libraries that are accessible within an Input or Output format (or otherwise outside of the standard transformation execution environment. This is necessary for reading directly out of HBase using the HBase TableInputFormat for example).

...

Pentaho MapReduce can be configured through the pentaho-mapreduceplugin.properties found in the plugin's base directory, or overridden per Pentaho MapReduce job entry if they are defined in the User Defined properties tab.

...

...

1. Unzip pentaho-mapreduce-libraries.zip, it contains a single lib/ directory with the required Kettle dependencies
2. Copy additional libraries to the lib/ directory
3. Zip up the lib/ directory into pentaho-mapreduce-libraries-custom.zip so the archive contains the lib/ with all jars within it (you may create subdirectories within lib/. All jars found in lib/ and its subdirectories will be added to the classpath of the executing job.)
4. Update pentaho-mapreduceplugin.properties and update the following properties:

   Code Block
   pmr.kettle.installation.id=custom pmr.libraries.archive.file=pentaho-mapreduce-libraries-custom.zip

The next time you execute Pentaho MapReduce the custom Kettle environment will be copied into HDFS at pmr.kettle.dfs.install.dir/custom and used when executing the job. You can switch between Kettle environments by specifying the pmr.kettle.installation.id property as a User Defined property per Pentaho MapReduce job entry or globally in the pentaho-mapreduceplugin.properties file*.

*Note: Only if the installation referenced by pmr.kettle.installation.id does not exist will the archive file and additional plugins currently configured will be used to "install" it into HDFS.

...

/opt/pentaho/mapreduce/
  +- 4.3.0/
  |   +- lib/
  |   |   +- kettle-core-{version}.jar
  |   |   +- kettle-engine-{version}.jar
  |   |   +- .. (Any files in the active Hadoop configuration's {{lib/pmr/}} directory)
  |   |   `- .. (all other required Kettle dependencies and optional jars - including JDBC drivers)
  |   `- plugins/
  |       +- pentaho-big-data-plugin/
  |       |  `- hadoop-configurations/
  |       |     `- hadoop-20/ (the active Hadoop configuration used to communicate with the cluster)
  |       |        +- lib/ (the {{lib/pmr/}} and {{lib/client/}} directories are omitted here)
  |       |        `- .. (all other jars)
  |       `- .. (additional optional plugins)
  `- custom/
      +- lib/
      |   +- kettle-core-{version}.jar
      |   +- kettle-engine-{version}.jar
      |   +- my-custom-code.jar
      |   `- .. (all other required Kettle dependencies and optional jars - including JDBC drivers)
      `- plugins/
          +- pentaho-big-data-plugin/
          |   ..
          `- my-custom-plugin/
              ..