2016-03-02 85 views

回答

3

默認情況下,所有的RDD都是懶惰的。直到你打電話,他們纔會真正做任何事情。所以不要添加懶惰,因爲這隻會延遲圍繞RDD創建元數據,而實際上並不影響執行。

val table = sparkContext.cassandraTable[Type](keyspace,tableName) 
val tableWithWhere = table.where("x = 5") 
val tableTransformed = table.map(x:Type => turnXIntoY(x)) 
//nothing has happened in C* or Spark on executors yet 
tableTransformed.collect // This causes spark to start doing work