Spark Streaming--Update

例如各大商城实时显示数据的案例：

package com.jiangnan.spark

import org.apache.spark.SparkConf
import org.apache.spark.streaming.{Seconds, StreamingContext}

object TestStreamUpdate extends App {
  //配置对象
  val conf = new SparkConf().setAppName("").setMaster("local[2]")
  //创建StreamContext
  val ssc = new StreamingContext(conf,Seconds(5))
  //重要：检查点目录的配置
  ssc.sparkContext.setCheckpointDir("E:\\cd\\data")
  //从socket接收数据
  val lineDStream = ssc.socketTextStream("master",8888)

  val words = lineDStream.flatMap(_.split(" ")).map(x=>(x,1))
  //使用updateStateByKey进行状态更新，统计从运行开始以来的单词总数
  val state = words.updateStateByKey[Int]((values:Seq[Int],state:Option[Int])=>{
    //values应该是新的数据
    val currentCount = values.foldLeft(0)(_+_)
    val stateCount = state.getOrElse(0)
    Some(currentCount+stateCount)
  })
  state.print()

  ssc.start()
  ssc.awaitTermination()

}