We propose a new framework identified as Spark that supports these apps although retaining the scalability and fault tolerance of MapReduce. To obtain these objectives, Spark introduces an abstraction called resilient dispersed datasets (RDDs). An RDD is actually a study-only collection of objects partitioned throughout a set of equipment that http://hotgraph46789.blog-gold.com/9633192/helping-the-others-realize-the-advantages-of-hotgraph