New to Claude Skills? Learn how to install them →

affaan-m on GitHub

Redis Patterns

Free

Best practices for using Redis in backend applications.

Get this skill

Free · Opens the source repo

What Redis Patterns does

Redis Patterns provides a comprehensive reference for implementing best practices with Redis in backend applications. It covers various data structures and their appropriate use cases, such as caching, session storage, rate limiting, and pub/sub messaging. This skill is particularly useful for developers and system architects who want to leverage Redis effectively in production environments.

The skill includes detailed explanations of core patterns such as Cache-Aside, Write-Through Cache, and Cache Invalidation, along with practical code examples in Python. It also addresses distributed locking mechanisms and rate limiting strategies, ensuring that users can manage concurrency and control access to resources efficiently. The provided Lua scripts and Python functions are designed to be easily integrated into existing applications, facilitating rapid implementation of these patterns.

Additionally, the skill outlines how to use Redis Streams for durable messaging and event processing, making it suitable for building scalable and responsive applications. By following the guidelines and examples in this skill, developers can enhance the performance and reliability of their applications while minimizing common pitfalls associated with Redis usage.

When to use it

Use this skill when you need to implement caching, rate limiting, or session management in your applications using Redis.

When not to use it

This skill may not be suitable for applications that do not utilize Redis or for users seeking advanced Redis configurations beyond the provided patterns.

What you can build with it

Implementing Caching

Use the Cache-Aside pattern to efficiently cache product data, reducing database load and improving response times.

Managing User Sessions

Utilize Redis for session storage, allowing quick access to user data and automatic expiration based on TTL.

Rate Limiting API Requests

Implement rate limiting using Redis to control the number of requests a user can make within a specified time window.

How to install Redis Patterns

View source

1. Install with the skills CLI

npx skills add affaan-m/ecc/redis-patterns --agent claude-code

2. Or install it manually

Download the skill folder and drop it into ~/.claude/skills/ for all projects, or .claude/skills/ to scope it to one repo. Restart Claude Code so it picks up the new skill.

Anthropic's agentic coding CLI, and the reference implementation of Agent Skills. Drop a skill folder into ~/.claude/skills and Claude Code loads it automatically whenever a task matches the skill's description. Claude Code docs

Inside SKILL.md

Written by affaan-m

Redis Patterns

一般的なバックエンド使用例に対するRedisベストプラクティスの参考資料。

How It Works

Redisはメモリ内データ構造ストアで、文字列、ハッシュ、リスト、セット、ソート済みセット、ストリームなどをサポートします。単一インスタンスでは個々のRedisコマンドは原子的ですが、マルチステップワークフローはLuaスクリプト、MULTI/EXECトランザクション、または明示的な同期化が必要です。RDBスナップショットまたはAOFログを通じてデータをオプションで永続化します。クライアントはRESPプロトコルを使用してTCP経由で通信します。接続プール不可欠でリクエストごとのハンドシェイクオーバーヘッドを回避します。

When to Activate

  • アプリケーションにキャッシング追加
  • レート制限またはスロットリング実装
  • 分散ロックまたはコーディネーション構築
  • セッションまたはトークンストレージ設定
  • Pub/SubまたはRedis Streams for messaging使用
  • 本番環境でRedis設定(プール、削除、クラスタリング)

Data Structure Cheat Sheet

Use CaseStructureExample Key
Simple cacheStringproduct:123
User sessionHashsession:abc
LeaderboardSorted Setscores:weekly
Unique visitorsSetvisitors:2024-01-01
Activity feedListfeed:user:456
Event streamStreamevents:orders
Counters / rate limitsString (INCR)ratelimit:user:123
Bloom filter / HLLHyperLogLoghll:pageviews

Core Patterns

Cache-Aside (Lazy Loading)

import redis
import json

r = redis.Redis(host='localhost', port=6379, decode_responses=True)

def get_product(product_id: int):
    cache_key = f"product:{product_id}"
    cached = r.get(cache_key)

    if cached:
        return json.loads(cached)

    product = db.query("SELECT * FROM products WHERE id = %s", product_id)
    r.setex(cache_key, 3600, json.dumps(product))  # TTL: 1 hour
    return product

Write-Through Cache

def update_product(product_id: int, data: dict):
    # DB書き込み先
    db.execute("UPDATE products SET ... WHERE id = %s", product_id)

    # キャッシュを即座に更新
    cache_key = f"product:{product_id}"
    r.setex(cache_key, 3600, json.dumps(data))

Cache Invalidation

# タグベース削除 — セット内で関連キーをグループ化
def cache_product(product_id: int, category_id: int, data: dict):
    key = f"product:{product_id}"
    tag = f"tag:category:{category_id}"
    pipe = r.pipeline(transaction=True)
    pipe.setex(key, 3600, json.dumps(data))
    pipe.sadd(tag, key)
    pipe.expire(tag, 3600)
    pipe.execute()

def invalidate_category(category_id: int):
    tag = f"tag:category:{category_id}"
    keys = r.smembers(tag)
    if keys:
        r.delete(*keys)
    r.delete(tag)

Session Storage

import time
import uuid

def create_session(user_id: int, ttl: int = 86400) -> str:
    session_id = str(uuid.uuid4())
    key = f"session:{session_id}"
    pipe = r.pipeline(transaction=True)
    pipe.hset(key, mapping={
        "user_id": user_id,
        "created_at": int(time.time()),
    })
    pipe.expire(key, ttl)
    pipe.execute()
    return session_id

def get_session(session_id: str) -> dict | None:
    data = r.hgetall(f"session:{session_id}")
    return data if data else None

def delete_session(session_id: str):
    r.delete(f"session:{session_id}")

Rate Limiting

Fixed Window (Simple)

def is_rate_limited(user_id: int, limit: int = 100, window: int = 60) -> bool:
    key = f"ratelimit:{user_id}:{int(time.time()) // window}"
    pipe = r.pipeline(transaction=True)
    pipe.incr(key)
    pipe.expire(key, window)
    count, _ = pipe.execute()
    return count > limit

Sliding Window (Lua — Atomic)

-- sliding_window.lua
local key = KEYS[1]
local now = tonumber(ARGV[1])
local window = tonumber(ARGV[2])
local limit = tonumber(ARGV[3])

redis.call('ZREMRANGEBYSCORE', key, 0, now - window)
local count = redis.call('ZCARD', key)

if count < limit then
    -- Use unique member (now + sequence) to avoid collisions within the same millisecond
    local seq_key = key .. ':seq'
    local seq = redis.call('INCR', seq_key)
    redis.call('EXPIRE', seq_key, math.ceil(window / 1000))
    redis.call('ZADD', key, now, now .. '-' .. seq)
    redis.call('EXPIRE', key, math.ceil(window / 1000))
    return 1
end
return 0
sliding_window = r.register_script(open('sliding_window.lua').read())

def allow_request(user_id: int) -> bool:
    key = f"ratelimit:sliding:{user_id}"
    now = int(time.time() * 1000)
    return bool(sliding_window(keys=[key], args=[now, 60000, 100]))

Distributed Locks

Distributed Lock (Single Node — SET NX PX)

import uuid

def acquire_lock(resource: str, ttl_ms: int = 5000) -> str | None:
    lock_key = f"lock:{resource}"
    token = str(uuid.uuid4())
    acquired = r.set(lock_key, token, px=ttl_ms, nx=True)
    return token if acquired else None

def release_lock(resource: str, token: str) -> bool:
    release_script = """
    if redis.call('get', KEYS[1]) == ARGV[1] then
        return redis.call('del', KEYS[1])
    else
        return 0
    end
    """
    result = r.eval(release_script, 1, f"lock:{resource}", token)
    return bool(result)

# Usage
token = acquire_lock("order:payment:123")
if token:
    try:
        process_payment()
    finally:
        release_lock("order:payment:123", token)

マルチノード設定の場合、フルRedlockアルゴリズムを実装する redlock-py ライブラリを使用してください。

Pub/Sub & Streams

Pub/Sub (Fire-and-Forget)

# Publisher
def publish_event(channel: str, payload: dict):
    r.publish(channel, json.dumps(payload))

# Subscriber (blocking — run in separate thread/process)
def subscribe_events(channel: str):
    pubsub = r.pubsub()
    pubsub.subscribe(channel)
    for message in pubsub.listen():
        if message['type'] == 'message':
            handle(json.loads(message['data']))

Redis Streams (Durable Queue)

# Producer
def emit(stream: str, event: dict):
    r.xadd(stream, event, maxlen=10000)  # Cap stream length

# Consumer group — guarantees at-least-once delivery
try:
    r.xgroup_create('events:orders', 'processor', id='0', mkstream=True)
except Exception:
    pass  # Group already exists

def consume(stream: str, group: str, consumer: str):
    while True:
        messages = r.xreadgroup(group, consumer, {stream: '>'}, count=10, block=2000)
        for _, entries in (messages or []):
            for msg_id, data in entries:
                process(data)
                r.xack(stream, group, msg_id)

配信保証、コンシューマーグループ、または再生が必要な場合、Pub/Sub代わりにStreamsを優先してください。

Key Design

Naming Conventions

# Pattern: resource:id:field
user:123:profile
order:456:status
cache:product:789

# Pattern: namespace:resource:id
myapp:session:abc123
myapp:ratelimit:user:123

# Pattern: resource:date (time-bound keys)
stats:pageviews:2024-01-01

TTL Strategy

Data TypeSuggested TTL
User session24h (86400)
API response cache5–15 min
Rate limit windowMatch window size
Short-lived tokens5–10 min
Leaderboard1h–24h
Static/reference data1h–1 week

常にTTLを設定してください。TTLなしのキーは無限に蓄積してメモリ圧力を引き起こします。

Connection Management

Connection Pooling

from redis import ConnectionPool, Redis

pool = ConnectionPool(
    host='localhost',
    port=6379,
    db=0,
    max_connections=20,
    decode_responses=True,
    socket_connect_timeout=2,
    socket_timeout=2,
)

r = Redis(connection_pool=pool)

Cluster Mode

from redis.cluster import RedisCluster

r = RedisCluster(
    startup_nodes=[{"host": "redis-1", "port": 6379}],
    decode_responses=True,
    skip_full_coverage_check=True,
)

Sentinel (High Availability)

from redis.sentinel import Sentinel

sentinel = Sentinel(
    [('sentinel-1', 26379), ('sentinel-2', 26379)],
    socket_timeout=0.5,
)
master = sentinel.master_for('mymaster', decode_responses=True)
replica = sentinel.slave_for('mymaster', decode_responses=True)

Eviction Policies

PolicyBehaviorBest For
noevictionError on write when fullQueues / critical data
allkeys-lruEvict least recently usedGeneral cache
volatile-lruLRU only among keys with TTLMixed data store
allkeys-lfuEvict least frequently usedSkewed access patterns
volatile-ttlEvict soonest-to-expirePrioritize long-lived data

redis.confを通じて設定:maxmemory-policy allkeys-lru

Anti-Patterns

Anti-PatternProblemFix
Keys with no TTLMemory grows unboundedAlways set TTL
KEYS * in productionBlocks the server (O(N))Use SCAN cursor
Storing large blobs (>100KB)Slow serialization, memory pressureStore reference + fetch from object store
Single Redis for everythingNo isolation between cache & queueUse separate DBs or instances
Ignoring connection pool limitsConnection exhaustion under loadSize pool to workload
Not handling cache miss stampedeThundering herd on cold startUse locks or probabilistic early expiry
FLUSHALL without thoughtWipes entire instanceScope deletes by key pattern

Cache Miss Stampede Prevention

import threading

_locks: dict[str, threading.Lock] = {}
_locks_mutex = threading.Lock()

def get_with_lock(key: str, fetch_fn, ttl: int = 300):
    cached = r.get(key)
    if cached:
        return json.loads(cached)

    with _locks_mutex:
        if key not in _locks:
            _locks[key] = threading.Lock()
        lock = _locks[key]
    with lock:
        cached = r.get(key)  # Re-check after acquiring lock
        if cached:
            return json.loads(cached)
        value = fetch_fn()
        r.setex(key, ttl, json.dumps(value))
        return value

マルチプロセスデプロイメント:インプロセスロックを上記の分散ロックセクション から acquire_lock/release_lock に置き換えてください。

Examples

Django/Flask APIエンドポイントにキャッシング追加: レスポンスに5分TTLでCache-asideを使用。リクエストパラメータでキーを指定。

ユーザーごとにAPIレート制限: 低トラフィックエンドポイントに固定ウィンドウを pipeline(transaction=True) で使用;正確なユーザーごと制限にはsliding-windowの Lua使用。

ワーカー間のバックグラウンドジョブ調整: 予想ジョブ期間を超えるTTLで acquire_lock を使用。常に finally ブロックでリリース。

複数購読者への通知のファンアウト: ファイアアンドフォーゲットにPub/Subを使用。保証配信または再生が必要な場合、Streamsに切り替え。

Quick Reference

PatternWhen to Use
Cache-asideRead-heavy, tolerate slight staleness
Write-throughStrong consistency required
Distributed lockPrevent concurrent access to a resource
Sliding window rate limitAccurate per-user throttling
Redis StreamsDurable event queue with consumer groups
Pub/SubBroadcast with no delivery guarantees needed
Sorted Set leaderboardRanked scoring, pagination
HyperLogLogApproximate unique count at low memory

Related

  • Skill: postgres-patterns — リレーショナルデータパターン
  • Skill: backend-patterns — APIおよびサービスレイヤーパターン
  • Skill: database-migrations — スキーマバージョニング
  • Skill: django-patterns — Djangoキャッシュフレームワーク統合
  • Agent: database-reviewer — 全データベースレビューワークフロー

Frequently asked questions about Redis Patterns

Similar skills