How compression works in nifi from kafka json

[#1]

How compression works in nifi
I have a consumercafcarerecord(jsontreereader+parquetrecordsetwriter)--mergecontent --putndfs(none compression) at my output, parquet files are read by spark, the problem is that the output from json is not large x2, but when I do snappy compression, the file namefile.parquet.snappy shrinks x5x6 at the output but the file cannot be opened by spark or it opens and the structure is no longer as good as it was at step 1 how to get the same structure from step 1 and x5x6 compression?