[Study notes]
A complete RDD mission consists of two parts: Transformation and Action. Transformation of the RDD is used to create, you can also put the old RDD to generate a new RDD through Transformation. For example, map transformation is a kind of operation, it has a function for each element of the incoming RDD a custom, and get a new element, and then all the new elements of a new RDD. But RDD feature is a big delay (lazy) calculation, that even if you perform these Transformation operations that are not performed. In other words, transformation action does not trigger the spark of program execution, they are only the first record of a series of operations RDD made only after a hit action operation, then all of the transformation will be performed in front. collect and reduce operations that action.
Reprinted from the original article: https://blog.csdn.net/qq_44596980/article/details/93308631