Spark hbase bulk load generates more than 15X data

Question

I have Spark dataframe with just 2 columns like { Key| Value}. And this dataframe has 10 million records. I am inserting this into HBase table (has 10 pre-split