Hi I have a requirement to dump parquet files in hive table using custom MR.
Parquet has so many data models- avro-parquet,proto-parquet,hive-parquet ? Which one is recommended over other for inmemory plain java objects. Hive internally uses MapredParquetOutputformat . Is it better than AvroParquetOutPutFormat ? I did n't find any sample program using MapredParquetOutputformat. What are pros and cons of MapredParquetOutputformat over AvroParquetOutPutFormat . Why ParquetOutputFormat needs a datamodel for converting java objects to parquet records ? Why not like ORCOutputFormat use a serde and convert any writable to ParquetRecord. Thanks
