Cassandra best practices when I need store json

Question

suppose that I have different customers, anyone that send me an different json structure ie:

customer_1: {'name': ..., 'surname': ...}

customer_2: {'name': ..., 'address': ..., 'amount': ...}

customer_3: {'type': ..., 'amount': ..., 'color': ..., 'light': ...}

[...]

Actually I use MongoDB to store and work with this data, but for some reason I need to migrate to Cassandra. What is the best practices to store this data?

Option 1: I create an column family / table for any customer so any customer has its table and data structure

Option 2: I store all in an table, and every time I have an new customer (with an own json) I make an alter table to create new column?

Option 3: I store the json in one table, as map type, but in this way I cannot work with the single value

Answer 1

Option 2 sounds the most compatible with Cassandra. One table can keep track of all the customers as rows, and each field of the JSON can be a column. When you access a specific customer, it will only spit back the columns in which it has data. Empty columns are virtually free in Cassandra.