apache spark - What is right way of Broadcasting a big Dataframe -
i have need broadcast 1 big file (converted dataframe) in spark 1.6.1 lookup. whether below code right way -
val file = sc.broadcast(dataframe_df)
then accessing using below code in udf.
def abc{ file.value.sqlcontext.sql(query) .... }
i read sc.broadcast brings data driver first , sends executors not approach this. right? help.
Comments
Post a Comment