df.write.parquet("AA_DWF_All.parquet",mode="overwrite") df_new = spark.read.parquet("AA_DWF_All.parquet") print(df_new.count())