Store Objects with Hashes

What you'll walk away with

Explain the core ideas behind Store Objects with Hashes
Run the sample Redis command or code and verify its output
Apply the technique correctly to the Tutorial Platform and production scenarios

Build the mental model

A Redis hash stores field/value pairs under one key. `HGET`, `HSET`, and `HINCRBY` update individual fields without parsing and replacing an entire JSON string. Nested objects, schema validation, or whole-object compression may favor another representation. TTL traditionally applies to the key, while field-expiry support depends on version, command availability, and design.

Connect it to a real scenario

Store a session profile as a hash and update plan, locale, and last_seen independently. Numeric fields can change atomically with `HINCRBY`. Very large hashes become single-key hotspots and can complicate migration and deletion latency.

Use hashes for field-level access
Use HINCRBY for atomic numeric updates
Bound hash size to avoid huge keys

Try the working example

shell

HSET session:abc123 user_id 42 locale mm plan free last_seen 2026-08-28T10:00:00Z
HGET session:abc123 user_id
HMGET session:abc123 locale plan
HINCRBY session:abc123 requests 1
HGETALL session:abc123
EXPIRE session:abc123 1800

You should see

You have a session hash with field-level updates and a 30-minute TTL.

5-minute try-it

Create a tutorial-card hash, increment views atomically, update the title independently, and set a TTL.

One important caution

Do not call `HGETALL` on an unbounded huge hash for every request. Fetch required fields and monitor cardinality.

Redis — Hashes — Redis

Easy traps

Do not call `HGETALL` on an unbounded huge hash for every request. Fetch required fields and monitor cardinality.
Validate sample commands on a local or test instance with recoverable data before applying them to production Redis.

Exercise

Create a tutorial-card hash, increment views atomically, update the title independently, and set a TTL.

You'll know it worked when: You have a session hash with field-level updates and a 30-minute TTL.