aggregate函數將每個分區裏面的元素進行聚合，然後用combine函數將每個分區的結果和初始值(zeroValue)進行combine操作。這個函數最終返回的類型不需要和RDD中元素類型一致。

scala> val rdd = sc.parallelize(List(18,28,7,66,-19,100,29,55,4),3)
rdd: org.apache.spark.rdd.RDD[Int] = ParallelCollectionRDD[2] at parallelize at 
 
<console>:24

//先對分區內的元素進行聚合；
scala> def InnerCom(a:Int, b:Int) : Int = {
     | println("InnerCom: " + a + " : " + b)
     | math.min(a,b)
     | }
InnerCom: (a: Int, b: Int)Int

//對聚合後的分區之間進行聚合
scala> def partitionCom(a:Int, b:Int): Int = {
     | println("partitionCom: " + a + " : " + b)
     | a + b
     | }
partitionCom: (a: Int, b: Int)Int

//3個分區，min(分區1)=7,min(分區2)=-19,min(分區1)=4
//50 + 7 + -19 + 4 = 42
scala> rdd.aggregate(50)(InnerCom,partitionCom)
InnerCom: 50 : 18
InnerCom: 18 : 28
InnerCom: 18 : 7
partitionCom: 50 : 7
InnerCom: 50 : 66
InnerCom: 50 : -19
InnerCom: -19 : 100
partitionCom: 57 : -19
InnerCom: 50 : 29
InnerCom: 29 : 55
InnerCom: 29 : 4
partitionCom: 38 : 4
res5: Int = 42
 

spark aggregate函數