I am just starting out with Cassandra and modeling my database. I've already have a de-normalized tables for MongoDB, and I want to model them to suit Cassandra.

Let's say that I have a Box and Items tables. Each box can contain various items (one to many). Each item have several field, including some that are incremental values that should be incremented +1 in a high rate. Each box belongs to a single user, a single user can have several boxes.

In Mongo I just combines the items collection inside the box items. I want to know what is the best way to model it I don't mind using any of the new features of Cassandra 1.2, because I intend to use it in production when the product launches this year.

Queries that I run against the DB:

- Paging trough the list of boxes with their items (ex. give me the 10 latest boxes with their items)
- Give me a box by its ID
- Insert a new item into a box
- increment an item value +1
- Show all the boxes that belongs to a specific user

I will really appreciate your help, because I am stuck at this and cannot continue developing my application.