I have cassandra cluster where CPU usage is between 10-20% normally with occcassional spikes to 40-50%. I am storing time series data ( which can come at times out of order