Check to see if there are any indices you don't need anymore. Check to see if you got indices on long text or string columns, and see if you can convert those to indices on bigints or int columns (by hashing the string).
In my opinion, it's not the amount of data that is the bigger issue, but the amount of reads/writes coming into the system. 100+ million rows with just 1 thread reading and writing to it really isn't a big deal, technically. If you have load issues, you might need to shard the data, do bulk inserts by buffering the data, increase the memory, cache commonly read rows in memory, etc.
Comments
Check to see if there are any indices you don't need anymore. Check to see if you got indices on long text or string columns, and see if you can convert those to indices on bigints or int columns (by hashing the string).
In my opinion, it's not the amount of data that is the bigger issue, but the amount of reads/writes coming into the system. 100+ million rows with just 1 thread reading and writing to it really isn't a big deal, technically. If you have load issues, you might need to shard the data, do bulk inserts by buffering the data, increase the memory, cache commonly read rows in memory, etc.