Reference · Cheat sheet

Data structures by complexity

Pick the type whose hot-path op is cheapest. Print it, pin it.

From lesson 0002See also glossary

Type → use → cost

Type	Reach for it when…	Cheap ops	O(N) traps
String	value, counter, flag, bitmap	`GET`/`SET`/`INCR` O(1)	—
Hash	a record with fields	`HGET`/`HSET` O(1)/field	`HGETALL`
List	queue / stack (touch the ends)	`LPUSH`/`RPUSH`/`LPOP`/`RPOP` O(1)	`LINDEX`, `LRANGE 0 -1`
Set	unique membership, tags	`SADD`/`SISMEMBER` O(1)	`SMEMBERS`, `SINTER` O(N·M)
Sorted set	ordered by a score	`ZSCORE` O(1) · `ZADD`/`ZRANK` O(log N) · `ZRANGE` O(log N + M)	—

The heuristic

Name your hot operation (the one run thousands×/sec). Choose the type that makes it O(1) or O(log N).
O(N) only on the rare path, or on collections you guarantee stay small.
Membership → set (SISMEMBER), never a scanned list.
Ordered-by-score → sorted set, never a re-sorted list.
Field access → hash, not one JSON string you read-modify-write whole.
Queue → list ends; reaching the middle means wrong structure.

Big-key rule

One key holding a giant collection = O(N) commands stall the shared thread and DEL is O(N). Bound collections, iterate with *SCAN, delete with UNLINK.

Encoding note

Small hashes/sets/zsets use a compact listpack (or intset) below config thresholds, so O(N) on them is trivial and memory-cheap. Redis auto-converts to hashtable/skiplist past the threshold. Inspect with OBJECT ENCODING key.

← lesson 0002 · Source: Redis Docs — Data types