We have set default MAX_FILESIZE=8GB on our cluster.
Main table has around 1100 columns. I create an index on one column. Max_filesize in secondary index table is set to 11.99MB ( i.e. (1/1100)*8GB) approximately .
This leads to excessive splitting of regions and a lot of regions for secondary index table.
So, i think, we need to set a lower bound on region size of secondary index tables in Phoenix.
I propose to have a Phoenix property(phoenix.global.index.maxfilesize.min) to set lower bound of max_filesize of secondary index table.
Please refer to this http://search-hadoop.com/m/9UY0h2DNmug2BAMko for details.