【大数据】广播变量调优

其实就是把一组数不需要再弄成rdd了，直接以变量形式播发给各个excuter，不使用广播变量那么会播发给各个task，excuter肯定比task少那么播发给excuter就优化一些了，这里相当于excuter变量相当于全局变量，每个task再使用时候向其最近的excuter申请就可以了，下面给出一个广播变量应用实例：

//查询日志中每个省所拥有的资源数


import org.apache.log4j.{Level, Logger}
import org.apache.spark.broadcast.Broadcast
import org.apache.spark.rdd.RDD
import org.apache.spark.{SparkConf, SparkContext}

/**
  *  广播变量的例子
  * Created by zhangjingcun on 2018/9/27 9:29.
  */
object IPLocation {
  val rulesFilePath = "D:\\data\\ip.txt"
  val accessFilePath = "D:\\data\\access.log"

  def main(args: Array[String]): Unit = {
    Logger.getLogger("org.apache.spark").setLevel(Level.OFF)

    val conf = new SparkConf().setAppName("IPLocation").setMaster("local[*]")
    val sc = new SparkContext(conf)

    //1：读取IP规则资源库
    val ipRulesLines: RDD[String] = sc.textFile(rulesFilePath)
    //2：整理IP规则
    //117.93.244.0|117.93.255.255|1969091584|1969094655|亚洲|中国|江苏|盐城||电信|320900|China|CN|120.139998|33.377631
    val ipRules: RDD[(Long, Long, String)] = ipRulesLines.map(line => {
      val fields = line.split("[|]")
      val startNum = fields(2).toLong
      val endNum = fields(3).toLong
      val province = fields(6)
      (startNum, endNum, province)
    })
    //
    //var result = ipRules.collect()
    //println(result.toBuffer)
    //3: 将IP规则收集到Driver（collect）
    val allIpRulesInDriver: Array[(Long, Long, String)] = ipRules.collect()
    //4：将全部的ip资源库通过广播的方式发送到Executor
    //广播之后，在Driver端获取了广播变量的引用（如果没有广播完，就不往下走）
    val broadcastRef: Broadcast[Array[(Long, Long, String)]] = sc.broadcast(allIpRulesInDriver)

    //5: 读取访问日志
    val accessLogLine: RDD[String] = sc.textFile(accessFilePath)
    //6: 整理访问日志
    //20090121000132095572000|125.213.100.123|show.51.com|/shoplist.php?phpfile=shoplist2.php&style=1&sex=137|Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; Mozilla/4.0(Compatible Mozilla/4.0(Compatible-EmbeddedWB 14.59 http://bsalsa.com/ EmbeddedWB- 14.59  from: http://bsalsa.com/ )|http://show.51.com/main.php|
    val provinceAddOne: RDD[(String, Int)] = accessLogLine.map(line=>{
      val fields = line.split("[|]")
      val ip = fields(1)
      val ipNum = MyUtils.ip2Long(ip)
      //通过广播变量的引用获取Executor中的全部IP规则，然后进行匹配ip规则
      val allIpRulesInExecutor: Array[(Long, Long, String)] = broadcastRef.value
      //根据规则进行查找，（用二分查找算法）
      var province = "未知"
      val index = MyUtils.binarySearch(allIpRulesInExecutor, ipNum)
      if(index != -1 ){
        province = allIpRulesInExecutor(index)._3
      }
      (province, 1)
    })

    //7： 按照省份的访问次数进行计数
    val reduceRDD: RDD[(String, Int)] = provinceAddOne.reduceByKey(_+_)

    //8：打印结果
    //    var result = reduceRDD.collect()
    //    println(result.toBuffer)

    //计算结果，将计算好的结果写入到mysql中
    //触发一个action，将数据写到mysql的逻辑函数传入
    //    reduceRDD.foreach(t =>{
    //      val conn = DriverManager.getConnection("jdbc:mysql://bigdata01:3306/bigdata", "root", "123456")
    //      val pstm = conn.prepareStatement("Insert Into .... values(?.?)")
    //      pstm.setString(1, t._1)
    //      pstm.setInt(2, t._2)
    //      pstm.executeUpdate()
    //      pstm.close()
    //      conn.close()
    //    })
    reduceRDD.foreachPartition(MyUtils.data2MySQL _)

    //9：释放资源
    sc.stop()
  }
}

例二：