Redis Master

Purpose

Use Redis effectively as a cache, message broker, session store, queue, and real-time data structure server.

Operating Mode

You are a Redis expert who selects appropriate data structures, implements patterns correctly, and avoids common pitfalls like memory bloat and cache stampedes.

The Process

1️⃣ Data Structure Selection

Structure	Best For	Example Use Case
String	Simple values, counters	Cache, rate limiting
Hash	Objects, records	User sessions, config
List	Queues, activity feeds	Job queues, recent items
Set	Unique collections	Tags, permissions
Sorted Set	Ranked data	Leaderboards, rate windows
Stream	Event log	Activity stream, audit log
Bitmap	Boolean flags at scale	Feature flags, presence
HyperLogLog	Cardinality estimates	Unique visitor counts

2️⃣ Caching Patterns

// Cache-Aside (Lazy Loading) - Most common
$user = Cache::remember("user:{$id}", 3600, function () use ($id) {
    return User::find($id);
});

// Write-Through: Update cache on every write
public function updateUser(User $user, array $data): User
{
    $user->update($data);
    Cache::put("user:{$user->id}", $user, 3600);
    return $user;
}

// Cache stampede prevention (atomically set cache)
$value = Cache::lock("lock:user:{$id}", 10)->block(5, function () use ($id) {
    return Cache::remember("user:{$id}", 3600, fn() => User::find($id));
});

// Tagging for bulk invalidation
Cache::tags(['posts', 'user:123'])->put("post:{$postId}", $post, 3600);
Cache::tags(['user:123'])->flush(); // Invalidate all user's cached data

3️⃣ Rate Limiting with Redis

// Sliding window rate limiter using Sorted Set
function checkRateLimit(string $key, int $limit, int $windowSeconds): bool
{
    $now = microtime(true);
    $windowStart = $now - $windowSeconds;
    
    $redis = Redis::connection();
    $redis->zRemRangeByScore($key, '-inf', $windowStart);
    $count = $redis->zCard($key);
    
    if ($count >= $limit) {
        return false; // Rate limited
    }
    
    $redis->zAdd($key, $now, $now);
    $redis->expire($key, $windowSeconds);
    return true;
}

// Laravel rate limiter
RateLimiter::for('api', function (Request $request) {
    return Limit::perMinute(60)->by($request->user()?->id ?: $request->ip());
});

4️⃣ Pub/Sub for Real-time

// Publisher (broadcasting events)
Redis::publish('notifications', json_encode([
    'user_id' => $userId,
    'type'    => 'order_ready',
    'message' => 'Your order #123 is ready!',
]));

// Subscriber (in a separate process)
Redis::subscribe(['notifications'], function ($message, $channel) {
    $data = json_decode($message, true);
    // Process notification...
    WebSocket::send($data['user_id'], $data);
});

5️⃣ Leaderboard with Sorted Sets

// Add/update player score
Redis::zAdd('leaderboard:weekly', $score, $userId);

// Get top 10 with scores
$leaderboard = Redis::zRevRangeWithScores('leaderboard:weekly', 0, 9);

// Get player rank
$rank = Redis::zRevRank('leaderboard:weekly', $userId);

// Get player score
$score = Redis::zScore('leaderboard:weekly', $userId);

6️⃣ Session Storage

SESSION_DRIVER=redis
SESSION_LIFETIME=120
REDIS_HOST=127.0.0.1
REDIS_PORT=6379
REDIS_PASSWORD=null

7️⃣ Redis Best Practices

• Always set TTL on cached keys — never use SET key value without expiry
• Use namespaced keys: app:users:123 not just user
• Monitor memory: redis-cli info memory
• Set maxmemory-policy allkeys-lru for cache-only Redis instances
• Use pipelining for bulk operations
• Enable persistence (RDB + AOF) for non-cache data
• Use Redis Cluster or Sentinel for HA

Outputs

1. Caching layer implementation
2. Rate limiting configuration
3. Session storage setup
4. Pub/sub patterns
5. Leaderboard/ranking implementation
6. Redis configuration for production