ETL with Hadoop - Big Data Made Easy: A Working Guide to the Complete Hadoop Toolset - page 292

To be able to show the functionality of PDI, I have created some tasks, including a Map Reduce job called “basic
mapred job” and the associated Map and Reduce tranformations called “mapper” and “reducer,” respectively. As you
can see, these are already open and displayed in Figure 10-3 . Also, note that the Explorer pane functionality on the left
of the figure changes depending on the current job or transformation that is displayed in the right pane.
The Explorer pane on the left of the window has View and Design tabs so that you can either see the content of
the task that you are working on or select functions to add to your job or transformation. The Working pane on the
right currently contains that Map Reduce job I just mentioned, called “basic mapred job” and the two transformations
called “mapper” and “reducer,” which it calls. If you drag the functionality icons from the Explorer pane to the
Working pane on the right, you can then configure and connect the functions to the flow of the task. (There will be
more about those functions later in the chapter.)
The Explorer view of the current mapper transformation, displayed in Figure 10-4 , shows that it contains Map
Reduce inputs and outputs, a filter, a field splitter, a set key, and a dummy task. (I'll explain all of these items in more
detail later.) The view also shows the task workflow hops that will be executed. As Figure 10-4 illustrates, you can
configure a slave server, meaning one instance of PDI uses another as a slave and gets that instance to run a task.
(To learn more about slave servers and how they work, see the Pentaho website at www.pentaho.com . )
Figure 10-4. Pentaho's Explorer view
Figure 10-5 provides a taste of the functionality that's available in the Design view. The two columns on the left
list the functions available for transformations (note the expanded big data section), while the two columns on the
right show the functions for jobs, again with an expanded big data view. You can construct transactions in a logical,
step-by-step manner by using these building blocks, then include the transactions in jobs that you schedule
within PDI.