Redis Strings, Lists, and Sets

SET, GET, LPUSH, SADD — the commands behind sessions, queues, and unique tags.

Intermediate12 min read

Read these first

Redis Architecture Overview

Why this matters

VaultCommerce product views use INCR for real-time counters, lists for audit trails, and sets for "customers who viewed this category."
Wrong type choice costs you: scanning a list for membership is O(n); a set gives O(1) lookups.
These three types cover most cache-aside and lightweight queue patterns before you reach sorted sets and streams.

StringSessions, counters

ListQueues, feeds

SetUnique tags

Sorted SetLeaderboards

HashUser profiles

StreamEvent log

Each type maps to a common backend pattern.

Strings: sessions and counters

Strings store text, integers, or binary up to 512 MB. Atomic increment/decrement makes them ideal for inventory and view counts without Postgres row locks.

Java

SET session:tok_a8f2 "{\"userId\":42,\"role\":\"buyer\"}" EX 1800
GET session:tok_a8f2
INCR product:SKU-8842:views
INCRBY inventory:SKU-8842 -1
SETNX lock:checkout:42 "worker-7" EX 30
MSET promo:FLASH10 "active" promo:WELCOME "active"
MGET promo:FLASH10 promo:WELCOME

String commands VaultCommerce uses

SET key value EX seconds — write with TTL (sessions, cache entries).
SETNX / SET ... NX — acquire a short-lived lock for idempotent checkout.
INCR / INCRBY — atomic counters for views, stock reservations, rate limits.
MGET — batch-read multiple product cache keys in one round trip.

Lists: recent activity feeds

Lists are linked lists with O(1) push/pop at head or tail. VaultCommerce stores the last ten cart events per user for support dashboards — not for random access by ID.

Java

LPUSH user:42:cart-events "added SKU-8842"
LPUSH user:42:cart-events "removed SKU-1100"
LRANGE user:42:cart-events 0 9
LLEN user:42:cart-events
LPOP user:42:cart-events

Use BLPOP when a worker blocks waiting for the next webhook retry job. Lists are queues, not indexes — avoid LINDEX in hot paths for large lists.

Sets: unique tags and audiences

Sets store unique unordered strings. O(1) add, remove, and membership checks make them perfect for deduplication.

Java

SADD category:electronics:viewers "user:42" "user:99" "user:42"
SISMEMBER category:electronics:viewers "user:42"
SCARD category:electronics:viewers
SINTER wishlist:42 wishlist:99
SUNION promo:email-list promo:sms-list

VaultCommerce tracks newsletter opt-ins with SADD newsletter:subscribers {email_hash} and checks duplicates before sending campaigns.

Python integration sketch

Java

import redis

r = redis.Redis(host="localhost", port=6379, decode_responses=True)

def cache_product(sku: str, payload: str, ttl: int = 3600) -> None:
    r.set(f"product:{sku}", payload, ex=ttl)

def reserve_stock(sku: str, qty: int = 1) -> int:
    remaining = r.decrby(f"inventory:{sku}", qty)
    if remaining < 0:
        r.incrby(f"inventory:{sku}", qty)  # rollback
        raise ValueError("out of stock")
    return remaining

Quick recall

Everything you need if you only revisit this box.

Strings: sessions, JSON cache, atomic counters — use SET EX, INCR, SETNX.
Lists: ordered queues and recent-event feeds — LPUSH, LRANGE, BLPOP.
Sets: unique membership and tag intersections — SADD, SISMEMBER, SINTER.
Match the type to access pattern: membership belongs in sets, not lists.
VaultCommerce prefixes keys by domain to keep namespaces isolated.
Batch reads with MGET to cut round trips on product grid pages.

Test yourself

Answer these before moving on — recall is what makes it stick.