Spark-1.2.0源码详读

以常用的几行代码为例，从源码中详细解读Spark的运行流程。

算法代码在spark shell中如下：

（1）val lines = sc.textFile("README.md")

（2）val words = lines.flatMap(x => x.split(" "))

（3）val wordCounts = words.map(x => (x, 1))

（4）val cacheCounts = wordCounts.cache()

（5）val reduced = cacheCounts.reduceByKey((a, b) => a + b)

（6）reduced.saveAsTextFile("haha")

（未完待续）