Using Map/Reduce AEs in SQL queries

The Netezza Analytics map/reduce framework provides the ability to invoke mappers and reducers directly from nzsql. There are three general Analytic Executables (AEs) registered with the mapreduce cartridge in the inza database. These can be used in a SQL query to execute a particular mapper, reducer, or combiner from a Java class. The AEs are registered as User-Defined Table Functions (UDTF) with the following names:

MapperAE
ReducerAE
CombinerAE

These AEs have a common signature:

<Task>AE(K1, … , Kn, V1, … , Vm, 'NZAE_PARAMETER2=<conf.xml>')

where:

K1, … , Kn: The key columns.
V1, … ,Vm: The value columns,
<conf.xml>: The path to a configuration file that contains all properties needed to run the task.

Properties

The following table lists the properties available to each AE.

Table 1.
Property	Mapper	Combiner	Reducer
mapreduce.job.run.dir	REQUIRED	YES	YES
mapreduce.job.jar	REQUIRED when job's jar available	REQUIRED when job's jar available	REQUIRED when job's jar available
mapreduce.job.map.class	REQUIRED
mapreduce.map.input.key.class	REQUIRED
mapreduce.map.input.value.class	REQUIRED
mapreduce.map.output.key.class	REQUIRED	REQUIRED	YES
mapreduce.map.output.key.columns.sizes	REQUIRED
mapreduce.map.output.value.class	REQUIRED	REQUIRED	YES
mapreduce.map.output.value.columns.sizes	REQUIRED
mapreduce.job.combine.class	REQUIRED	REQUIRED
mapreduce.combine.output.key.class	REQUIRED	REQUIRED	REQUIRED when mapreduce.job.combine.class is defined
mapreduce.combine.output.key.columns.sizes	REQUIRED	REQUIRED
mapreduce.combine.output.value.class	REQUIRED	REQUIRED	REQUIRED when mapreduce.job.combine.class is defined
mapreduce.combine.output.value.columns.sizes		REQUIRED
mapreduce.job.reduce.class			REQUIRED
mapreduce.reduce.output.key.class			REQUIRED
mapreduce.reduce.output.key.columns.sizes			REQUIRED
mapreduce.reduce.output.value.class			REQUIRED
mapreduce.reduce.output.value.columns.sizes			REQUIRED