Redis Sorted Sets (Part 2)
This lesson covers advanced sorted set features such as score-range queries and set operations.
ZRANGEBYSCORE: Query by Score Range
ZRANGEBYSCORE retrieves elements based on a score range.
Basic Usage
REDIS
ZADD leaderboard 100 "player1" 150 "player2" 200 "player3" 250 "player4" 300 "player5"
# Get elements with scores between 150 and 250
ZRANGEBYSCORE leaderboard 150 250
1) "player2"
2) "player3"
3) "player4"
# Get elements with scores
ZRANGEBYSCORE leaderboard 150 250 WITHSCORES
1) "player2"
2) "150"
3) "player3"
4) "200"
5) "player4"
6) "250"
Using -inf and +inf
REDIS
# Get elements with score <= 200
ZRANGEBYSCORE leaderboard -inf 200
1) "player1"
2) "player2"
3) "player3"
# Get elements with score >= 200
ZRANGEBYSCORE leaderboard 200 +inf
1) "player3"
2) "player4"
3) "player5"
# Get all elements
ZRANGEBYSCORE leaderboard -inf +inf
Open and Closed Intervals
REDIS
# Default is closed interval [150, 250]
ZRANGEBYSCORE leaderboard 150 250
# Open interval (150, 250) — excludes 150 and 250
ZRANGEBYSCORE leaderboard (150 (250
# Half-open [150, 250)
ZRANGEBYSCORE leaderboard 150 (250
# Half-open (150, 250]
ZRANGEBYSCORE leaderboard (150 250
💡 Syntax: Use parentheses
( to indicate an open interval (exclusive), no parentheses for a closed interval (inclusive).
LIMIT Pagination
REDIS
# Get elements scored 100-300, starting from index 1, take 2
ZRANGEBYSCORE leaderboard 100 300 LIMIT 1 2
1) "player2"
2) "player3"
ZREVRANGEBYSCORE: Query by Score Range (High to Low)
REDIS
# Get elements scored 150-250, from high to low
ZREVRANGEBYSCORE leaderboard 250 150 WITHSCORES
1) "player4"
2) "250"
3) "player3"
4) "200"
5) "player2"
6) "150"
⚠️ Note: The score range in ZREVRANGEBYSCORE goes from highest to lowest (250 150), not low to high.
ZRANGEBYLEX: Query by Lexicographical Order
When all elements have the same score, you can query by lexicographical (alphabetical) order.
Basic Usage
REDIS
# All elements have score 0
ZADD myzset 0 "apple" 0 "banana" 0 "cherry" 0 "date"
# Get all elements in lexicographical order
ZRANGEBYLEX myzset - +
1) "apple"
2) "banana"
3) "cherry"
4) "date"
# Get elements in lexicographical range [banana, cherry]
ZRANGEBYLEX myzset [banana [cherry
1) "banana"
2) "cherry"
# Get elements starting with "b"
ZRANGEBYLEX myzset [b (c
1) "banana"
💡 Use case: ZRANGEBYLEX is ideal for implementing autocomplete and dictionary lookups.
ZREMRANGEBYRANK: Remove by Rank Range
REDIS
ZADD leaderboard 100 "p1" 150 "p2" 200 "p3" 250 "p4" 300 "p5"
# Remove elements with rank 0 to 1 (lowest 2 scores)
ZREMRANGEBYRANK leaderboard 0 1
(integer) 2
ZRANGE leaderboard 0 -1
1) "p3"
2) "p4"
3) "p5"
ZREMRANGEBYSCORE: Remove by Score Range
REDIS
ZADD leaderboard 100 "p1" 150 "p2" 200 "p3" 250 "p4" 300 "p5"
# Remove elements with scores between 100 and 200
ZREMRANGEBYSCORE leaderboard 100 200
(integer) 3
ZRANGE leaderboard 0 -1 WITHSCORES
1) "p4"
2) "250"
3) "p5"
4) "300"
Use Case: Clean Up Expired Data
REDIS
# Delayed queue (score is timestamp)
ZADD delay:queue 1719129600 "task:1"
ZADD delay:queue 1719133200 "task:2"
# Remove completed tasks (timestamp less than current time)
ZREMRANGEBYSCORE delay:queue -inf 1719129600
ZREMRANGEBYLEX: Remove by Lexicographical Range
REDIS
ZADD myzset 0 "apple" 0 "banana" 0 "cherry" 0 "date"
# Remove elements in lexicographical range [banana, cherry]
ZREMRANGEBYLEX myzset [banana [cherry
(integer) 2
ZRANGE myzset 0 -1
1) "apple"
2) "date"
Sorted Set Operations
ZUNIONSTORE: Union
REDIS
# Create two sorted sets
ZADD zset1 1 "a" 2 "b" 3 "c"
ZADD zset2 4 "b" 5 "c" 6 "d"
# Calculate union (default: sum scores)
ZUNIONSTORE result 2 zset1 zset2
(integer) 4
ZRANGE result 0 -1 WITHSCORES
1) "a"
2) "1"
3) "d"
4) "6"
5) "b"
6) "6" # 2 + 4
7) "c"
8) "8" # 3 + 5
ZUNIONSTORE with Weights
REDIS
# Set weights
ZUNIONSTORE result 2 zset1 zset2 WEIGHTS 2 3
# zset1 scores multiplied by 2, zset2 scores multiplied by 3
ZRANGE result 0 -1 WITHSCORES
1) "a"
2) "2" # 1 * 2
3) "d"
4) "18" # 6 * 3
5) "b"
6) "16" # 2*2 + 4*3
7) "c"
8) "21" # 3*2 + 5*3
ZUNIONSTORE Aggregate Functions
REDIS
# Use MIN aggregate (take the minimum score)
ZUNIONSTORE result 2 zset1 zset2 AGGREGATE MIN
# Use MAX aggregate (take the maximum score)
ZUNIONSTORE result 2 zset1 zset2 AGGREGATE MAX
# Use SUM aggregate (sum, default)
ZUNIONSTORE result 2 zset1 zset2 AGGREGATE SUM
ZINTERSTORE: Intersection
REDIS
ZADD zset1 1 "a" 2 "b" 3 "c"
ZADD zset2 4 "b" 5 "c" 6 "d"
# Calculate intersection
ZINTERSTORE result 2 zset1 zset2
(integer) 2
ZRANGE result 0 -1 WITHSCORES
1) "b"
2) "6" # 2 + 4
3) "c"
4) "8" # 3 + 5
ZDIFFSTORE: Difference (Redis 6.2+)
REDIS
ZADD zset1 1 "a" 2 "b" 3 "c"
ZADD zset2 4 "b" 5 "c" 6 "d"
# Calculate difference
ZDIFFSTORE result 2 zset1 zset2
(integer) 1
ZRANGE result 0 -1 WITHSCORES
1) "a"
2) "1"
ZMSCORE: Batch Get Scores (Redis 6.2+)
REDIS
ZADD leaderboard 100 "player1" 200 "player2" 300 "player3"
ZMSCORE leaderboard "player1" "player2" "player4"
1) "100"
2) "200"
3) (nil) # player4 does not exist
ZPOPMAX and ZPOPMIN: Pop Highest/Lowest Score Elements
ZPOPMAX: Pop the Highest-Scoring Element
REDIS
ZADD leaderboard 100 "p1" 200 "p2" 300 "p3"
# Pop the highest-scoring element
ZPOPMAX leaderboard
1) "p3"
2) "300"
# Pop 2 highest-scoring elements
ZPOPMAX leaderboard 2
1) "p2"
2) "200"
3) "p1"
4) "100"
ZPOPMIN: Pop the Lowest-Scoring Element
REDIS
ZADD leaderboard 100 "p1" 200 "p2" 300 "p3"
ZPOPMIN leaderboard
1) "p1"
2) "100"
BZPOPMAX and BZPOPMIN: Blocking Pop
REDIS
# If the sorted set is empty, block and wait up to 10 seconds
BZPOPMAX leaderboard 10
# Listen on multiple sorted sets
BZPOPMAX zset1 zset2 zset3 10
💡 Use case: BZPOPMAX/BZPOPMIN can be used to implement priority queues.
Advanced Sorted Set Applications
Use Case 1: Sliding Window Rate Limiting
REDIS
# Record user request timestamps
ZADD ratelimit:user:1 1719129600 "req1"
ZADD ratelimit:user:1 1719129605 "req2"
ZADD ratelimit:user:1 1719129610 "req3"
# Count requests in the last 60 seconds
ZCOUNT ratelimit:user:1 1719129540 1719129600
# Clean up requests older than 60 seconds
ZREMRANGEBYSCORE ratelimit:user:1 -inf 1719129540
Use Case 2: Autocomplete
REDIS
# All words with score 0
ZADD autocomplete 0 "apple" 0 "application" 0 "appetite" 0 "banana"
# Query words starting with "app"
ZRANGEBYLEX autocomplete [app [app\xff
1) "apple"
2) "application"
3) "appetite"
Use Case 3: Timeline
REDIS
# User timeline (score is timestamp)
ZADD timeline:user:1 1719129600 "post:1"
ZADD timeline:user:1 1719129605 "post:2"
ZADD timeline:user:1 1719129610 "post:3"
# Get the latest 10 posts
ZREVRANGE timeline:user:1 0 9
# Get posts within a time range
ZRANGEBYSCORE timeline:user:1 1719129600 1719129610
Use Case 4: Multi-Dimensional Ranking
REDIS
# Product sales ranking
ZADD sales:rank 1000 "product:1"
ZADD sales:rank 2000 "product:2"
# Product rating ranking
ZADD rating:rank 4.5 "product:1"
ZADD rating:rank 4.8 "product:2"
# Combined ranking (sales weight 0.6, rating weight 0.4)
ZUNIONSTORE combined:rank 2 sales:rank rating:rank WEIGHTS 0.6 0.4
❓ FAQ
Q What's the difference between ZRANGEBYSCORE and ZRANGE?
A ZRANGE queries by rank range; ZRANGEBYSCORE queries by score range.
Q How do I get the top N elements by score?
A Use
ZREVRANGE key 0 N-1 WITHSCORES.Q How do weights work in ZUNIONSTORE?
A Weights are multipliers. WEIGHTS 2 3 means multiply scores in the first set by 2 and the second set by 3.
Q Does ZRANGEBYLEX require all elements to have the same score?
A Yes. ZRANGEBYLEX assumes all elements have the same score and sorts them lexicographically.
Q How do I implement a priority queue?
A Use ZADD to add tasks (score = priority), and ZPOPMIN to pop the highest-priority task.
📖 Summary
- ZRANGEBYSCORE queries by score range, supports open intervals and LIMIT
- ZRANGEBYLEX queries by lexicographical order (all same score)
- ZREMRANGEBYRANK/SCORE/LEX removes elements by range
- ZUNIONSTORE/ZINTERSTORE set operations with weights and aggregates
- ZPOPMAX/ZPOPMIN pops highest/lowest scoring elements
- BZPOPMAX/BZPOPMIN blocking pop for priority queues
- Advanced applications: sliding window rate limiting, autocomplete, timeline, multi-dimensional ranking
📝 Exercises
- Score range query: Use ZRANGEBYSCORE to query elements in different score ranges
- Range deletion: Use ZREMRANGEBYSCORE to clean up expired data
- Set operations: Use ZUNIONSTORE to merge multiple leaderboards with different weights
- Sliding window: Use a sorted set to implement sliding window rate limiting
Next Lesson
In the next lesson, we will learn Redis HyperLogLog, covering cardinality estimation.