Insert overwrite vs Delete HDFS partition and write parquet file in spark

问题

Is there any major performance issue for parquet in using

hiveContext.sql("INSERT OVERWRITE TABLE table_name PARTITION(dt='2016-06-08', country) , select x,y,z,country from DFTmpTable");

VS

DeleteHDFSfile(/table/dt='2016-06-08')
DF.write().mode("append").partitionBy("dt","country").parquet("/table/dt='2016-06-08'")

来源：https://stackoverflow.com/questions/39810659/insert-overwrite-vs-delete-hdfs-partition-and-write-parquet-file-in-spark