Hello,

elgg implement as core matadatastring table where all data from community are stored. Problem of this access is very complicated devided for partition table because only id is in table. Problem is I/O read write too. I have real experience with th last project where we have 300 millions rows in one table where without partition our project was down. ideal partition devided is time scale but community life provide many of records in each time and query is compose give me data for all groups or all discuss. Create partition for all group is not possible because only 256 partition subtable can be for on table.  

I thinking about cluster where table can be in memory but problem with all data in one segment is still alive.     

Have you any suggest how extend ellg core table to provide partition devided. Thank you for any idea.