Module Review: CQL

You’ve mastered the query language that powers the world’s most scalable databases. It looks like SQL, but requires a completely different mindset.

1. Key Takeaways

  • Schema is Deployment: A Keyspace defines replication strategies (NetworkTopologyStrategy vs SimpleStrategy). A Table defines data distribution (Partition Key).
  • Primary Key Anatomy: PRIMARY KEY ((Partition Key), Clustering Key).
  • Partition Key: Determines WHERE data lives (which node).
  • Clustering Key: Determines ORDER of data on disk.
  • The Golden Rule: Always provide the Partition Key in your WHERE clause.
  • ALLOW FILTERING: The keyword of death. It causes a full cluster scan. Never use in production.
  • Writes are Cheap: INSERT/UPDATE/DELETE are all appended to the Commit Log and MemTable. No read-before-write.
  • Collections: Use Set, List, Map for nested data. Use frozen<> for performance or Primary Key usage.

2. Interactive Flashcards

Test your recall.

Loading...
Answer
1 / 5

3. Cheat Sheet

Concept Command / Syntax Notes
Create Keyspace CREATE KEYSPACE k WITH replication = {'class': 'NetworkTopologyStrategy', 'dc1': 3}; Always use NTS in prod.
Create Table CREATE TABLE t (id uuid, val text, PRIMARY KEY(id)); Simple PK.
Composite PK PRIMARY KEY ((part_key), clust_key) Groups by part_key, sorts by clust_key.
Insert / Update INSERT INTO t (id, val) VALUES (...) Same as Update (Upsert).
TTL USING TTL 86400 Expires data after N seconds.
Select SELECT * FROM t WHERE id = ? Must allow coordinator to find the node.
Frozen UDT frozen<my_type> Serialized as one blob. Required for PKs.
Batch BEGIN BATCH ... APPLY BATCH Use for atomicity across tables, NOT performance.

4. Next Steps

Now that you can model and query data, it’s time to understand Consistency Levels.