Redis Data Types Overview

Introduction

Redis is not a dumb string cache. It offers strings, lists, sets, sorted sets, hashes, streams, bitmaps, HyperLogLog, and geospatial indexes — each optimized for specific access patterns.

Picking the wrong type wastes memory and forces O(N) commands where O(1) was available. A JSON string for a session object prevents partial field updates; a set when you need ranking misses ZREVRANGE.

This overview maps business problems to Redis types so you choose correctly before writing code.

Understanding the topic

Key concepts

• String — binary-safe blob, counters (INCR), cache entries.
• Hash — field map for objects (session, profile).
• List — ordered sequence, queue (LPUSH/BRPOP).
• Set — unique unordered members, tags, graph edges.
• Sorted set — member + score, leaderboards, delayed queues.
• Stream — append-only log with consumer groups (Kafka-lite).
• Bitmap / HLL / GEO — specialized analytics and location.

flowchart LR
  Cache[String] --> API
  Session[Hash] --> Auth
  Queue[List] --> Worker
  Tags[Set] --> Filter
  Rank[ZSet] --> Board
  Events[Stream] --> Consumer

Step-by-step explanation

1. Every key has exactly one type; TYPE command returns it.
2. Wrong-type commands error (WRONGTYPE Operation against a key).
3. Redis converts encodings internally as collections grow/shrink.
4. OBJECT ENCODING reveals internal representation.
5. MEMORY USAGE estimates bytes for capacity planning.

Syntax reference

Common commands

• TYPE before operating on unknown keys.
• MEMORY USAGE — plan shard sizing.
• WRONGTYPE means you picked the wrong API for that key.

TYPE mykey
OBJECT ENCODING mykey
MEMORY USAGE mykey

# Quick reference
# String: SET GET INCR
# Hash: HSET HGET
# List: LPUSH BRPOP
# Set: SADD SISMEMBER
# ZSet: ZADD ZRANGE
# Stream: XADD XREADGROUP

Informative example

Same user modeled three ways — only hash fits partial updates efficiently:

# String blob — rewrite entire JSON on one field change
SET user:1 '{"name":"Ada","plan":"pro"}'

# Hash — update one field
HSET user:1 plan enterprise lastLogin "2026-07-01"

# Wrong: Set loses field semantics
SADD user:1 name:Ada plan:pro

Use hash when you read/update individual fields. String JSON is OK for read-mostly cache blobs.

Real-world use

Real-world use cases

• String — API response cache, feature flag value.
• Hash — shopping cart line attributes, user session.
• List — lightweight job queue between services.
• Set — unique visitors per article, mutual followers.
• Sorted set — game leaderboard, priority task queue.
• Stream — order events, audit trail with consumer groups.

Best practices

• Match type to read/write pattern, not ORM habit.
• Run TYPE in runbooks when debugging legacy keys.
• Prefer hash over JSON string for mutable objects.
• Use streams over pub/sub when you need retention.
• Document type conventions in team key naming guide.
• Sample MEMORY USAGE on representative keys before launch.

Common mistakes

• Everything stored as JSON strings.
• Using list as set (duplicates allowed).
• Sorted set when set suffices (extra memory for scores).
• Pub/sub when streams needed for replay.

Advanced interview questions

Summary

Redis types are purpose-built — not interchangeable.