spark-sql性能优化之——多线程实现多Job并发执行

直接上代码

val spark = SparkSession.builder()
    .appName("name")
    .master("local[2]")
    .getOrCreate()

  val df = spark.read.json("src\\main\\resources\\json.txt")

  df.show()

  //没有多线程处理的情况，连续执行两个Action操作，生成两个Job
  df.rdd.saveAsTextFile("")
  df.rdd.saveAsTextFile("")

  //用Executor实现多线程方式处理Job
  val dfList = Array(df,df)
  val executorService = Executors.newFixedThreadPool()
  for(df <- dfList) {
    executorService.submit(new Callable[Boolean]() {
      def call() : Boolean  = {
        df.show()
        true
      }
    })
  }

  executorService.shutdown()

  spark.stop()