redis-advanced-types

Redis Advanced Types Skill

Bitmaps

Memory-efficient bit operations for flags and counters. 512MB max = 2^32 bits.

Bitmap Commands

SETBIT key offset value           # Set bit - O(1)
GETBIT key offset                 # Get bit - O(1)
BITCOUNT key [start end [BYTE|BIT]]  # Count 1s - O(N)
BITOP AND|OR|XOR|NOT destkey key [key ...]  # Bitwise ops - O(N)
BITPOS key bit [start [end [BYTE|BIT]]]     # Find position - O(N)
BITFIELD key [GET type offset] [SET type offset value] [INCRBY type offset increment]

Pattern 1: User Activity Tracking

# Track daily logins (1 bit per day)
SETBIT user:123:logins:2024 0 1   # Jan 1
SETBIT user:123:logins:2024 4 1   # Jan 5
SETBIT user:123:logins:2024 9 1   # Jan 10

# Count login days in January
BITCOUNT user:123:logins:2024 0 3  # First 4 bytes = 31 days

# Check specific day
GETBIT user:123:logins:2024 4  # Returns 1

Pattern 2: Feature Flags

# Bit positions represent features
# 0=dark_mode, 1=notifications, 2=beta_features
SETBIT user:123:features 0 1  # Enable dark mode
SETBIT user:123:features 2 1  # Enable beta

# Check feature
GETBIT user:123:features 0  # 1 = enabled

Pattern 3: Daily Active Users

# Track DAU with bitmap
SETBIT dau:2024-01-15 123 1  # User 123 active
SETBIT dau:2024-01-15 456 1  # User 456 active

# Count DAU
BITCOUNT dau:2024-01-15

# Weekly active (OR across days)
BITOP OR wau:2024-w3 dau:2024-01-15 dau:2024-01-16 dau:2024-01-17
BITCOUNT wau:2024-w3

# Users active all 7 days (AND)
BITOP AND daily-active:week dau:2024-01-15 dau:2024-01-16 ...

HyperLogLog

Probabilistic cardinality estimation with 0.81% standard error. Only 12KB per key!

HyperLogLog Commands

PFADD key element [element ...]    # Add - O(1)
PFCOUNT key [key ...]              # Count - O(1) per key
PFMERGE destkey sourcekey [sourcekey ...]  # Merge - O(N)

Pattern 1: Unique Visitors

# Track hourly unique visitors
PFADD visitors:2024-01-15:10 "user:1" "user:2" "user:3"
PFADD visitors:2024-01-15:11 "user:2" "user:4"

# Count unique for hour
PFCOUNT visitors:2024-01-15:10  # ~3

# Daily unique (merge hours)
PFMERGE visitors:2024-01-15 visitors:2024-01-15:10 visitors:2024-01-15:11
PFCOUNT visitors:2024-01-15  # ~4

Pattern 2: Unique Search Queries

PFADD searches:2024-01 "redis tutorial" "redis commands"
PFCOUNT searches:2024-01

Streams

Append-only log with consumer groups for reliable message processing.

Stream Commands

# Writing
XADD stream [NOMKSTREAM] [MAXLEN|MINID [=|~] threshold] *|id field value [field value ...]
XLEN stream                        # Get length - O(1)

# Reading
XREAD [COUNT count] [BLOCK ms] STREAMS stream [stream ...] id [id ...]
XRANGE stream start end [COUNT count]
XREVRANGE stream end start [COUNT count]

# Consumer Groups
XGROUP CREATE stream group id|$ [MKSTREAM] [ENTRIESREAD n]
XREADGROUP GROUP group consumer [COUNT count] [BLOCK ms] [NOACK] STREAMS stream [stream ...] id [id ...]
XACK stream group id [id ...]
XPENDING stream group [[IDLE min-idle-time] start end count [consumer]]
XCLAIM stream group consumer min-idle-time id [id ...]
XAUTOCLAIM stream group consumer min-idle-time start [COUNT count]

# Management
XTRIM stream MAXLEN|MINID [=|~] threshold
XDEL stream id [id ...]
XINFO STREAM|GROUPS|CONSUMERS stream [group]

Pattern 1: Event Sourcing

# Add events
XADD orders:events * event_type "created" order_id "123" amount "99.99"
XADD orders:events * event_type "paid" order_id "123" payment_id "pay_456"
XADD orders:events * event_type "shipped" order_id "123" tracking "TRK789"

# Read all events for replay
XRANGE orders:events - +

# Read new events only
XREAD BLOCK 5000 STREAMS orders:events $

Pattern 2: Consumer Group Processing

# Create consumer group
XGROUP CREATE orders:events processors $ MKSTREAM

# Consumer 1 reads
XREADGROUP GROUP processors consumer-1 COUNT 10 BLOCK 5000 STREAMS orders:events >

# Acknowledge processed
XACK orders:events processors 1704067200000-0

# Check pending (unacked)
XPENDING orders:events processors

# Claim stale messages (consumer died)
XAUTOCLAIM orders:events processors consumer-2 60000 0-0 COUNT 10

Pattern 3: Capped Stream (Logs)

# Add with auto-trim
XADD logs MAXLEN ~ 10000 * level "error" message "Connection failed"

# Manual trim
XTRIM logs MAXLEN ~ 10000

Geospatial

Location-based data with distance and radius queries.

Geo Commands

GEOADD key [NX|XX] [CH] longitude latitude member [longitude latitude member ...]
GEOPOS key member [member ...]
GEODIST key member1 member2 [M|KM|FT|MI]
GEOHASH key member [member ...]
GEOSEARCH key FROMMEMBER member|FROMLONLAT lon lat BYRADIUS radius M|KM|FT|MI|BYBOX width height M|KM|FT|MI [ASC|DESC] [COUNT count [ANY]] [WITHCOORD] [WITHDIST] [WITHHASH]
GEOSEARCHSTORE dest src FROMMEMBER member|FROMLONLAT lon lat BYRADIUS radius unit|BYBOX width height unit [ASC|DESC] [COUNT count [ANY]] [STOREDIST]

Pattern 1: Store Locator

# Add store locations
GEOADD stores -122.4194 37.7749 "store:sf"
GEOADD stores -118.2437 34.0522 "store:la"
GEOADD stores -73.9857 40.7484 "store:nyc"

# Find stores within 50km
GEOSEARCH stores FROMLONLAT -122.4 37.8 BYRADIUS 50 km WITHCOORD WITHDIST

# Distance between stores
GEODIST stores "store:sf" "store:la" km  # ~559 km

Pattern 2: Nearby Users

# Update user location
GEOADD users:location -122.4194 37.7749 "user:123"

# Find nearby users
GEOSEARCH users:location FROMMEMBER "user:123" BYRADIUS 5 km COUNT 10 WITHDIST

# Store results for caching
GEOSEARCHSTORE nearby:user:123 users:location FROMMEMBER "user:123" BYRADIUS 5 km

Pattern 3: Delivery Zone Check

# Define delivery zones (store + radius)
GEOADD zones -122.4194 37.7749 "zone:downtown"

# Check if address in delivery zone
GEODIST zones "zone:downtown" "customer:addr" km
# If < 10, can deliver

Command Complexity

Assets

• stream-consumer.py - Stream consumer implementation
• geo-search.lua - Atomic geo operations

References

• ADVANCED_TYPES_GUIDE.md - Complete guide

Troubleshooting Guide

Common Issues & Solutions

1. Bitmap Memory Explosion

Cause: Setting high offset creates sparse bitmap

SETBIT sparse 100000000 1  # Creates ~12MB

Fix: Use hash with bitmap segments

# Instead of SETBIT user:flags 1000000 1
HSET user:flags segment:1000 "\x01"

2. HyperLogLog Accuracy

Issue: Count varies on repeated PFCOUNT

Expected: 0.81% standard error is normal

# For 1M unique elements
# Expected range: 991,900 - 1,008,100

3. Stream Consumer Lag

Diagnosis:

XINFO GROUPS mystream
# Check lag field

Fix:

• Add more consumers
• Increase batch size
• Check consumer processing time

4. Pending Messages Growing

XPENDING mystream mygroup

Cause: Consumers crashing without ACK

Fix:

# Claim and reprocess
XAUTOCLAIM mystream mygroup new-consumer 300000 0-0 COUNT 100

5. Geo Precision Issues

Issue: Geohash precision ~0.6m max

Fix: Store exact coords in hash if needed

HSET location:123 lat "37.7749000000" lon "-122.4194000000"

Debug Checklist

□ Data type correct? (TYPE key)
□ Bitmap offset reasonable? (< 2^32)
□ Stream consumer group exists? (XINFO GROUPS)
□ Pending messages cleared? (XPENDING)
□ HLL merged correctly? (PFMERGE)
□ Geo coordinates valid? (-180 to 180, -85.05 to 85.05)

Error Codes Reference

Test Template

# test_redis_advanced.py
import redis
import pytest
import time

@pytest.fixture
def r():
    return redis.Redis(decode_responses=True)

class TestBitmaps:
    def test_user_activity(self, r):
        r.delete("test:activity")
        r.setbit("test:activity", 0, 1)
        r.setbit("test:activity", 5, 1)
        assert r.bitcount("test:activity") == 2
        assert r.getbit("test:activity", 0) == 1
        assert r.getbit("test:activity", 1) == 0
        r.delete("test:activity")

class TestHyperLogLog:
    def test_unique_count(self, r):
        r.delete("test:hll")
        r.pfadd("test:hll", "a", "b", "c", "a")  # 'a' duplicate
        count = r.pfcount("test:hll")
        assert 2 <= count <= 4  # Allow HLL variance
        r.delete("test:hll")

    def test_merge(self, r):
        r.delete("test:hll1", "test:hll2", "test:merged")
        r.pfadd("test:hll1", "a", "b")
        r.pfadd("test:hll2", "b", "c")
        r.pfmerge("test:merged", "test:hll1", "test:hll2")
        count = r.pfcount("test:merged")
        assert 2 <= count <= 4
        r.delete("test:hll1", "test:hll2", "test:merged")

class TestStreams:
    def test_basic_stream(self, r):
        r.delete("test:stream")
        msg_id = r.xadd("test:stream", {"event": "test"})
        assert msg_id is not None

        messages = r.xrange("test:stream", "-", "+")
        assert len(messages) == 1
        assert messages[0][1]["event"] == "test"
        r.delete("test:stream")

    def test_consumer_group(self, r):
      

---

*Content truncated.*