Cassandra is designed to work as a big-data in cluster nodes. Cassandra design is focused in fast writes (like a event/visit/… log) and at the same time to offer all the capacities of replication and fail tolerance. It has eventual consistency and has operations of map/reduce for complex queries.
Cassandra is oriented to be used as a replacement of a traditional database. Its infrastructure allows to replicate automatically the database across many data-centers around the world automatically, and at the same time, it can use itself as a backup. The main objective is the performance and data reliability.