参考：https://www.jianshu.com/p/ce737d2fc8fc

    Receiver提高吞吐方式：

    • 提高并行度：修改 block time，分区数 = batch time / block time

    • 提高batch time读取kafka的量(前提是kafka吞吐足够)：通过创建多个streaming，然后再 union

      1. JavaPairDStream<String, String> streaming = null;
      2. if (numPartitions > 1) {
      3.   List<JavaPairDStream<String, String>> kafkaStreams = new ArrayList(numPartitions);
      5.   for(int i = 0; i < receiverNum; ++i) {
      6.       kafkaStreams.add(KafkaUtilsExt.createStream(jssc, kafkaParams, topics, StorageLevel.MEMORY_AND_DISK_SER()));
      7.   }
      9.   streaming = jssc.union((JavaPairDStream)kafkaStreams.get(0), kafkaStreams.subList(1, kafkaStreams.size()));
      10. } else {
      11.   streaming = KafkaUtilsExt.createStream(jssc, kafkaParams, topics, StorageLevel.MEMORY_AND_DISK_SER());
      12. }