Spark SQL 4

1) mock data
2) DataFrame vs Dataset
DataFrame = Dataset[Row]
SchemaRDD ============> DataFrame ==========> Dataset
rename due to compile-time type safety(编译时就能提示错误)
OO structure change 


Catalog API    对元数据的管理：库，表，列，cache  cache有几种不同方式
 
 
SQLConf    最权威的API 比官网文档还权威




select /*+ mapjoin(A)*/


 
 
 
ThriftServer/HiveServer2 ==> Spark Application(SparkContext)
beeline
jdbc


*****


Hive on Spark


SparkTask


SQL==> MapTask/ReduceTask
   ==> SparkTask