What is a wide-column store and how does Cassandra's data model work?

Expected Interview Answer

Step-by-Step Explanation

1. Step 1

   Create a keyspace

   Define a keyspace with a replication strategy and factor — it is the top-level container for tables.

2. Step 2

   Design tables around queries

   List the exact queries the app runs first, then build one table per access pattern rather than normalizing.

3. Step 3

   Choose the primary key

   Pick a partition key that spreads data evenly and clustering columns that order rows within a partition.

4. Step 4

   Denormalize deliberately

   Duplicate data across tables so each read hits a single partition, trading storage for read speed.

5. Step 5

   Insert and query by partition

   Write rows and read them by supplying the full partition key so requests avoid cross-node scatter.

Code Example

CREATE KEYSPACE store
  WITH replication = {'class': 'SimpleStrategy', 'replication_factor': 3};

CREATE TABLE store.orders_by_customer (
  customer_id uuid,
  order_id timeuuid,
  total decimal,
  status text,
  PRIMARY KEY (customer_id, order_id)
) WITH CLUSTERING ORDER BY (order_id DESC);

INSERT INTO store.orders_by_customer (customer_id, order_id, total, status)
  VALUES (now(), now(), 49.99, 'shipped');

Follow-up Questions

• How does Cassandra decide which node stores a given partition?
• Why does Cassandra discourage secondary indexes and joins?
• What is the difference between a keyspace and a table?
• How does denormalization affect storage and consistency?