MapR Data Generator

Data generator for MapR Data Platform

How to build

mvn clean compile assembly:single     

This should give you /mapr-data-gen-1.0.jar in your target folder.

How to run

./bin/spark-submit --master yarn \ 
--class class com.github.anicolaspp.Generator \ 
mapr-data-gen-1.0.jar [OPTIONS]

Current options are:

  usage: parquet-generator
   -C,--compress <arg>                          <String> compression type, valid values are:
                                                uncompressed, snappy, gzip,
                                                lzo (default: uncompressed)
   -f,--format <arg>                            <String> output format type (e.g., parquet, maprdb (default), etc.)
   -o,--output <arg>                            <String> the output file name (default: /ParqGenOutput.parquet)
   -O,--options <arg>                           <str,str> key,value strings that will be passed to the data source of spark in
                                                writing. E.g., for parquet you may want to re-consider parquet.block.size. The
                                                default is 128MB (the HDFS block size).
   -p,--partitions <arg>                        <int> number of output partitions (default: 1)
   -r,--rows <arg>                              <long> total number of rows (default: 10)
   -R,--rangeInt <arg>                          <int> maximum int value, value for any Int column will be generated between
                                                [0,rangeInt), (default: 2147483647)
   -s,--size <arg>                              <int> any variable payload size, string or payload in IntPayload (default: 100)
   -S,--show <arg>                              <int> show <int> number of rows (default: 0, zero means do not show)
   -t,--tasks <arg>                             <int> number of tasks to generate this data (default: 1)

An example run would be :

./bin/spark-submit --master yarn \
--class com.github.anicolaspp.Generator mapr-data-gen-1.0.jar  \
-o /user/mapr/tables/test_gen -r 84 -s 42 -p 12 -f maprdb

This will create 984 ( = 12 * 84) rows for case class Data as [String, Int, Array[Byte], Double, Float, Long, String] with 42 bytes byte array and 42 chars String, and save this as a MapR-DB table in /user/mapr/tables/test_gen.

We can generate parquet data in the following way.

./bin/spark-submit --master yarn \
--class com.github.anicolaspp.Generator mapr-data-gen-1.0.jar  \
-o /user/mapr/data/parquet -r 84 -s 42 -p 12 -f parquet 

In this case the same data is generated and saved in /user/mapr/data/parquet

We can generate stream data in the following way.

./spark-submit --master yarn \
--deploy-mode client \
--num-executors 12 \
--executor-cores 2 \
--executor-memory 5G \
--class com.github.anicolaspp.Generator \
~/mapr-data-gen-1.0.jar -o /user/mapr/streams/random_data:t1 -r 1000000 -s 1024 -p 24 -f mapres -c 50 -t 20

1. Notice that -o points to a MapR Stream and includes the topic (t1 in our case).
2. We are generating 1000000 rows
3. We use -t 20 to indicate that we use 20 tasks to generate the data
4. We use -p 24 to indicate that we use 24 partitions to write to the stream
5. We use -c 50 to indicate that we use 50 threads on each partition to write to MapR-ES.

Finally, if we need to run the stream data generator for a period of time (we want to test a consumer while writing data) we can add -m argument to indicate the time we want to run it (in minutes).

./spark-submit --master yarn \
--deploy-mode client \
--num-executors 12 \
--executor-cores 2 \
--executor-memory 5G \
--class com.github.anicolaspp.Generator \
~/mapr-data-gen-1.0.jar -o /user/mapr/streams/random_data:t1 -r 10000 -s 1024 -p 5 -f mapres -c 50 -t 100 -m 5

Notice the same arguments than before, but we have added -m 5 to indicate that we want to write the generated data during 5 minutes. The application will stop after that.