Partition unbounded collection like Kafka source

[wang Wu]

Hi,
Supposed that a Kafka topic has 3 partitions only. Now we want to partition it 
into 20 partition, each one will produce an output collection. The purpose is 
to write to the sink in parallel from all 20 output collections.

Will this code achieve that purpose?

 KafkaIO.Read<byte[], Long> reader =
      KafkaIO.<byte[], Long>read()
          .withConsumerFactoryFn(
              new ConsumerFactoryFn(
                  topic, 10, numElements, OffsetResetStrategy.EARLIEST)) // 10 
partitions

  PCollection<Long> input = 
p.apply(reader.withoutMetadata()).apply(KafkaToCassandraRow).apply(CassadraIO.write);

Regards
Dinh