I have a postgres table which is not partitioned and it contains close to 1.5 billion records. It has about 200 customer_id values but the no. of records for each customer v