SCAN — Iterating Keys

SCAN vs KEYS

KEYS pattern is simple but dangerous in production: it blocks the entire Redis server while it scans all keys. For a database with millions of keys, this can take seconds and freeze your application.

SCAN solves this by iterating keys in batches without blocking:

SCAN — Cursor-Based Iteration

SCAN cursor [MATCH pattern] [COUNT count]

• cursor — start with 0, use the returned cursor for the next call
• Returns 0 as the cursor when the full iteration is complete
• MATCH — filter keys by pattern (applied after fetching, not before)
• COUNT — hint for how many keys to check per call (not a guarantee)

Example — iterate all keys:

SCAN 0

Output:

1) "0"        -- next cursor (0 means scan complete)
2) 1) "key1"
   2) "key2"
   3) "key3"

Example — iterate with pattern:

SET user:1 "Alice"
SET user:2 "Bob"
SET product:1 "Laptop"

SCAN 0 MATCH user:*

Output:

1) "0"
2) 1) "user:1"
   2) "user:2"

How Cursor Iteration Works

In real Redis, SCAN returns a cursor that you use in subsequent calls:

cursor = 0
do:
  cursor, keys = SCAN cursor MATCH pattern COUNT 100
  process(keys)
while cursor != 0

This emulator always returns cursor 0 (scan complete in one call), matching real Redis behavior for small datasets.

HSCAN, SSCAN, ZSCAN

These work similarly but iterate over hash fields, set members, and sorted set members:

HSET user:1 name "Alice" email "a@b.com" age "30"
HSCAN user:1 0

SADD tags "redis" "cache" "db"
SSCAN tags 0

ZADD scores 100 "Alice" 90 "Bob"
ZSCAN scores 0

Production Pattern

In application code with ioredis or redis-py:

# Python: iterate all keys matching pattern
for key in redis.scan_iter("user:*"):
    print(key)

Your Task

Set three keys: session:abc → "user1", session:def → "user2", cache:page1 → "html". Use SCAN with MATCH to find only session keys.