Redis Advanced Data Types: Bitmap, Geo and Stream

This is article 41 in the Big Data series. Introduces three Redis advanced data types: Bitmap, Geo, and Stream.

Complete illustrated version: CSDN Original | Juejin

Bitmap

Core Concept: Bitmap is a data structure using binary bits, each bit can only be 0 or 1, used to represent element value or state.

Space Efficiency

Bitmap’s biggest advantage is extremely high space efficiency. Store 10,000 users’ sign-in status:

• Traditional (boolean in Set): ~40KB
• Bitmap: ~1.25KB (saves over 97%)

Common Use Cases

1. User sign-in/check-in
2. Online user statistics
3. Bloom filter
4. Permission management
5. Deduplication counting

Core Commands

Command	Description
SETBIT key offset value	Set bit value at specified offset (0 or 1)
GETBIT key offset	Get bit value at specified offset
BITCOUNT key [start end]	Count bits with value 1
BITOP operation destkey key [key ...]	Bitwise operations on multiple Bitmaps (AND/OR/XOR/NOT)

Practical Example: User Sign-in

# User 1000 signs in on 2024-01-01
SETBIT user:sign:1000 20240101 1

# Check if user 1000 signed in on 2024-01-01
GETBIT user:sign:1000 20240101

# Count total sign-in days for user 1000
BITCOUNT user:sign:1000

---

Geo Type

Redis Geo type internally uses Sorted Set, combining three core technologies: Z-order curve, Base32 encoding, and GeoHash algorithm.

Z-Order Curve

Z-order curve is a space-filling curve that maps multi-dimensional data to one-dimensional while preserving spatial locality—points close in geographic location are also close in one-dimensional encoding.

Encoding Process (example: coordinate (3, 5)):

1. Convert X=3, Y=5 to binary: 011, 101
2. Interleave bits (Y first then X): 011011
3. Convert to decimal: 27

Base32 Encoding

Base32 uses 32 characters (0-9, b-z minus confusing characters) to convert binary data to printable ASCII:

1. Split binary data into 5-bit groups
2. Map each group (0-31) to Base32 alphabet
3. Pad with = if less than 5 bits at the end

GeoHash Algorithm

GeoHash encodes any location on Earth into a compact string. Redis internally uses 52-bit integer stored in Sorted Set’s score, achieving efficient nearby queries through z-score sorting.

Core Commands

# Add geographic location (longitude latitude member name)
GEOADD user:addr 111.11 44.44 ww 112.22 43.33 kk

# Get GeoHash encoding
GEOHASH user:addr ww

# Get coordinates
GEOPOS user:addr ww

# Calculate distance between two points (unit: km)
GEODIST user:addr ww kk km

# Find nearby members (Redis 6.2+)
GEOSEARCH user:addr FROMMEMBER ww BYRADIUS 100 km ASC

---

Stream Type

Stream is a data structure introduced in Redis 5.0, providing persistent message queue functionality, filling the gap that Pub/Sub cannot persist.

Main Capabilities

• Auto-generate message IDs (format: millisecond timestamp-sequence number)
• Message traversal and range query
• Blocking/non-blocking read
• Consumer Group support
• Pending message acknowledgment mechanism
• Queue monitoring and message backtracking

Core Commands

# Write message (* means auto-generate ID)
XADD topic:001 * name wzk age 18

# Read range messages (- means minimum ID, + means maximum ID)
XRANGE topic:001 - +

# Consume message (start from ID=0, read 1 message)
XREAD COUNT 1 STREAMS topic:001 0

# Create consumer group
XGROUP CREATE topic:001 group1 0

# Consumer group read
XREADGROUP GROUP group1 consumer1 COUNT 1 STREAMS topic:001 >

# Acknowledge message
XACK topic:001 group1 <message-id>

Difference from Pub/Sub

Feature	Stream	Pub/Sub
Message Persistence	Supported	Not supported
Consumer Acknowledgment	Supported (ACK)	Not supported
Consumer Group	Supported	Not supported
Historical Messages	Can retrieve	Cannot retrieve

Stream is suitable for scenarios requiring reliable message delivery, while Pub/Sub is more suitable for real-time broadcasting where message loss is acceptable.