版权声明:本博客都是作者10多年工作总结 https://blog.csdn.net/Peter_Changyb/article/details/85235082
import org.apache.spark.{SparkConf, SparkContext}
/**
* Created by cyb on 2018/9/14.
*/
object WordCount {
def main(args: Array[String]) {
//非常重要,是通向Spark集群的入口
val conf = new SparkConf().setAppName("WC")
.setJars(Array("C:\\HelloSpark\\target\\hello-spark-1.0.jar"))
.setMaster("spark://node-1.itcast.cn:7077")
val sc = new SparkContext(conf)
//textFile会产生两个RDD:HadoopRDD -> MapPartitinsRDD
sc.textFile(args(0)).cache()
// 产生一个RDD :MapPartitinsRDD
.flatMap(_.split(" "))
//产生一个RDD MapPartitionsRDD
.map((_, 1))
//产生一个RDD ShuffledRDD
.reduceByKey(_+_)
//产生一个RDD: mapPartitions
.saveAsTextFile(args(1))
sc.stop()
}
}
原理解析: