Tell me about Cassandra: consistency, guarantees, features

Apache Cassandra is a distributed NoSQL database focused on high availability and scalability.

Consistency: Cassandra uses a tunable consistency model. You can choose the consistency level for read and write operations: from one node (ONE) to all nodes (ALL). This allows balancing between speed and strict consistency.

Guarantees:

• Eventual consistency by default, but strong consistency can be achieved by configuring read and write levels.
• Writes are idempotent, with a timestamp-based conflict resolution mechanism.
• Supports data replication with strategies like SimpleStrategy and NetworkTopologyStrategy.

Features:

• Column-oriented data model with dynamic columns.
• No rigid schema, providing flexibility.
• High fault tolerance through replication and automatic recovery.
• Uses Gossip protocol for state sharing between nodes.
• Supports CQL (Cassandra Query Language), similar to SQL.

Example of setting the consistency level in Java driver during a write:

session.execute(
  new SimpleStatement("INSERT INTO users (id, name) VALUES (?, ?)", id, name)
    .setConsistencyLevel(ConsistencyLevel.QUORUM));

Here, QUORUM means the write must be acknowledged by the majority of replicas.