The Model
Cassandra is a distributed wide-column database built for high write throughput and high availability.
Primary Key
The primary key is split into a partition key and a clustering key. The partition key decides where the data lives; the clustering key decides the order inside that partition.
Consistency
Cassandra lets you choose consistency per query. ONE, QUORUM, and ALL are the standard levels to know.
Write Path
Writes go through the commit log and memtable, then flush into SSTables. That is why Cassandra is fast for write-heavy workloads.
Modeling
Model around queries, not normalized entity design. Cassandra is most useful when you know your access patterns up front.
Final Shape
The clean interview answer is: Cassandra is a leaderless, AP-biased, wide-column store optimized for write-heavy workloads with clear query patterns.