MatterAI Documentation | AI Super Intelligence for Engineering

Redis Anti-Patterns Overview

Using Redis as a Primary Database

// Anti-pattern: Using Redis as the only database for all data
// Storing all user data in Redis
redis.hset('user:1000', 'username', 'john_doe');
redis.hset('user:1000', 'email', 'john@example.com');
redis.hset('user:1000', 'profile', JSON.stringify(largeProfileObject));
redis.hset('user:1000', 'preferences', JSON.stringify(userPreferences));
redis.hset('user:1000', 'activity_history', JSON.stringify(extensiveActivityHistory));

// Better approach: Use Redis as a cache with a primary database
// Store core data in a primary database (e.g., PostgreSQL)
const user = await db.query('SELECT * FROM users WHERE id = $1', [1000]);

// Cache frequently accessed data in Redis
redis.hset('user:1000', 'username', user.username);
redis.hset('user:1000', 'email', user.email);
redis.expire('user:1000', 3600); // Set expiration time (1 hour)

// For large objects, consider separate cache keys with their own TTLs
redis.set('user:1000:profile', JSON.stringify(user.profile));
redis.expire('user:1000:profile', 7200); // 2 hours

Using Redis as your only database can lead to data loss, as Redis is primarily an in-memory store. Use Redis as a cache or for specific use cases (like session storage, rate limiting) alongside a persistent primary database.

Not Setting Key Expiration

// Anti-pattern: Not setting expiration on keys
redis.set('session:user:1000', sessionData);
redis.set('cache:product:5', productData);
redis.set('temp:import:progress', '75%');

// Better approach: Set appropriate TTLs for different types of data
// Session data (expire after 24 hours)
redis.set('session:user:1000', sessionData);
redis.expire('session:user:1000', 86400);

// Or in a single command
redis.set('session:user:1000', sessionData, 'EX', 86400);

// Cache data (expire after 1 hour)
redis.set('cache:product:5', productData, 'EX', 3600);

// Temporary processing data (expire after 10 minutes)
redis.set('temp:import:progress', '75%', 'EX', 600);

Not setting expiration times on keys can lead to memory exhaustion as Redis accumulates stale data. Set appropriate TTLs (Time To Live) for different types of data based on how long they should be kept.

Using Keys Command in Production

// Anti-pattern: Using KEYS command in production
redis.keys('user:*', (err, keys) => {
  keys.forEach(key => {
    redis.get(key, (err, value) => {
      // Process each user
    });
  });
});

// Better approach: Use SCAN for iterating over keys
async function processAllUsers() {
  let cursor = '0';
  do {
    const [nextCursor, keys] = await redis.scan(
      cursor, 
      'MATCH', 
      'user:*',
      'COUNT',
      100
    );
    cursor = nextCursor;
    
    // Process batch of keys
    for (const key of keys) {
      const value = await redis.get(key);
      // Process each user
    }
  } while (cursor !== '0');
}

The KEYS command blocks the Redis server until it completes, which can cause performance issues in production. Use SCAN instead, which iterates through keys incrementally without blocking the server.

Storing Large Objects

// Anti-pattern: Storing large objects in Redis
const largeObject = {
  // A very large object with many nested properties
  // Could be several MB in size
};

redis.set('large:object', JSON.stringify(largeObject));

// Better approach: Break down large objects or use a different storage solution
// Option 1: Store only the necessary fields
redis.hset('object:1000', 'id', largeObject.id);
redis.hset('object:1000', 'name', largeObject.name);
redis.hset('object:1000', 'summary', largeObject.summary);

// Option 2: Store a reference to the object in another database
redis.set('object:1000:ref', 'stored_in_s3_bucket_x');

// Option 3: If you must store the whole object, consider compression
const compressed = await compressData(JSON.stringify(largeObject));
redis.set('large:object:compressed', compressed);

Storing large objects in Redis can lead to increased memory usage and poor performance. Break down large objects into smaller pieces, store only the necessary fields, or use a different storage solution for large data.

Not Using the Right Data Structure

// Anti-pattern: Using the wrong data structure
// Storing a list of user IDs as a string
redis.set('team:members', '1,2,3,4,5');

// To add a new member, we need to get, modify, and set again
redis.get('team:members', (err, members) => {
  const memberArray = members.split(',');
  memberArray.push('6');
  redis.set('team:members', memberArray.join(','));
});

// Better approach: Use the appropriate data structure
// Using a Set for team members
redis.sadd('team:members', '1', '2', '3', '4', '5');

// Adding a new member is atomic and efficient
redis.sadd('team:members', '6');

// Checking if a user is a member is also efficient
redis.sismember('team:members', '3', (err, isMember) => {
  // isMember will be 1 if user 3 is a member, 0 otherwise
});

Using the wrong data structure for your use case can lead to inefficient operations and complex application code. Choose the appropriate Redis data structure (string, list, set, sorted set, hash) based on how you need to access and manipulate the data.

Not Using Pipelining or Transactions

// Anti-pattern: Multiple individual commands
redis.set('user:1000:name', 'John');
redis.set('user:1000:email', 'john@example.com');
redis.set('user:1000:age', '30');
redis.sadd('users', '1000');

// Better approach: Use pipelining
const pipeline = redis.pipeline();
pipeline.set('user:1000:name', 'John');
pipeline.set('user:1000:email', 'john@example.com');
pipeline.set('user:1000:age', '30');
pipeline.sadd('users', '1000');
pipeline.exec((err, results) => {
  // Handle results
});

// For operations that need to be atomic, use transactions
const multi = redis.multi();
multi.set('user:1000:name', 'John');
multi.set('user:1000:email', 'john@example.com');
multi.set('user:1000:age', '30');
multi.sadd('users', '1000');
multi.exec((err, results) => {
  // Handle results
});

Sending individual commands to Redis creates network overhead. Use pipelining to send multiple commands in a single network round trip, and transactions (MULTI/EXEC) when you need operations to be executed atomically.

Not Implementing Proper Error Handling

// Anti-pattern: No error handling
redis.set('key', 'value');
redis.get('key', (err, value) => {
  // No error handling
  processValue(value);
});

// Better approach: Implement proper error handling
redis.set('key', 'value', (err) => {
  if (err) {
    console.error('Error setting key:', err);
    // Implement retry logic or fallback mechanism
    return;
  }
  console.log('Key set successfully');
});

redis.get('key', (err, value) => {
  if (err) {
    console.error('Error getting key:', err);
    // Implement retry logic or fallback mechanism
    return;
  }
  if (value === null) {
    console.log('Key not found');
    // Handle missing key scenario
    return;
  }
  processValue(value);
});

Not handling Redis errors can lead to application crashes or incorrect behavior. Implement proper error handling for all Redis operations, including connection errors, command errors, and handling of null values.

Not Monitoring Redis Memory Usage

// Anti-pattern: Not monitoring Redis memory usage
// Just using Redis without monitoring

// Better approach: Regularly monitor memory usage
// Set up a scheduled task to check Redis memory usage
async function monitorRedisMemory() {
  const info = await redis.info('memory');
  const usedMemory = parseInt(info.match(/used_memory:(\d+)/)[1]);
  const usedMemoryHuman = info.match(/used_memory_human:([\w.]+)/)[1];
  const maxMemory = parseInt(info.match(/maxmemory:(\d+)/)[1]);
  const maxMemoryHuman = info.match(/maxmemory_human:([\w.]+)/)[1];
  
  const memoryUsagePercent = (usedMemory / maxMemory) * 100;
  
  console.log(`Redis memory usage: ${usedMemoryHuman} / ${maxMemoryHuman} (${memoryUsagePercent.toFixed(2)}%)`);
  
  // Alert if memory usage is above threshold
  if (memoryUsagePercent > 80) {
    sendAlert(`Redis memory usage is high: ${memoryUsagePercent.toFixed(2)}%`);
  }
}

// Run every 5 minutes
setInterval(monitorRedisMemory, 5 * 60 * 1000);

Not monitoring Redis memory usage can lead to unexpected out-of-memory errors or performance degradation. Set up monitoring for Redis memory usage and implement alerts for when memory usage exceeds certain thresholds.

Not Configuring Eviction Policies

# Anti-pattern: Using default configuration or no eviction policy
# redis.conf with default settings (noeviction)
# maxmemory 0
# maxmemory-policy noeviction

# Better approach: Configure appropriate maxmemory and eviction policy
# redis.conf with configured memory limit and eviction policy
maxmemory 2gb
maxmemory-policy allkeys-lru

Not configuring appropriate eviction policies can lead to Redis running out of memory and crashing. Configure maxmemory and an appropriate eviction policy (allkeys-lru, volatile-lru, etc.) based on your use case.

Using Redis Pub/Sub for Critical Messages

// Anti-pattern: Using Redis Pub/Sub for critical messages
// Publisher
redis.publish('orders', JSON.stringify(newOrder));

// Subscriber
const subscriber = redis.duplicate();
subscriber.subscribe('orders');
subscriber.on('message', (channel, message) => {
  const order = JSON.parse(message);
  processOrder(order);
});

// Better approach: Use a message queue with persistence and acknowledgment
// Using a dedicated message broker like RabbitMQ, Kafka, or Redis Streams

// With Redis Streams (for Redis 5.0+)
// Producer
redis.xadd('orders', '*', 'order', JSON.stringify(newOrder));

// Consumer group (with persistence and acknowledgment)
redis.xgroup('CREATE', 'orders', 'order-processors', '$', 'MKSTREAM');

// Consumer
async function processOrders() {
  while (true) {
    try {
      const results = await redis.xreadgroup(
        'GROUP', 'order-processors', 'consumer-1',
        'COUNT', 10,
        'BLOCK', 2000,
        'STREAMS', 'orders', '>'
      );
      
      if (results) {
        for (const [stream, messages] of results) {
          for (const [id, [_, orderStr]] of messages) {
            const order = JSON.parse(orderStr);
            try {
              await processOrder(order);
              // Acknowledge successful processing
              await redis.xack('orders', 'order-processors', id);
            } catch (err) {
              console.error(`Error processing order ${id}:`, err);
              // Don't acknowledge, will be redelivered to another consumer
            }
          }
        }
      }
    } catch (err) {
      console.error('Error reading from stream:', err);
      await new Promise(resolve => setTimeout(resolve, 1000));
    }
  }
}

processOrders();

Redis Pub/Sub doesn’t persist messages or provide delivery guarantees, making it unsuitable for critical messages. Use a proper message queue with persistence and acknowledgment for critical messages, such as Redis Streams, RabbitMQ, or Kafka.

Not Using Connection Pooling

// Anti-pattern: Creating new connections for each operation
function getUserData(userId) {
  const redis = new Redis();
  return redis.get(`user:${userId}`)
    .then(data => {
      redis.disconnect();
      return data;
    })
    .catch(err => {
      redis.disconnect();
      throw err;
    });
}

// Better approach: Use connection pooling
// Initialize a connection pool once
const redisPool = new GenericPool({
  create: () => new Redis(),
  destroy: client => client.quit(),
  max: 10, // Maximum number of connections in the pool
  min: 2   // Minimum number of connections to keep in the pool
});

// Use a connection from the pool for each operation
async function getUserData(userId) {
  const redis = await redisPool.acquire();
  try {
    return await redis.get(`user:${userId}`);
  } finally {
    await redisPool.release(redis);
  }
}

Creating new Redis connections for each operation is inefficient and can lead to connection limits being reached. Use connection pooling to reuse connections across operations.

Not Using Redis Cluster for High Availability

// Anti-pattern: Using a single Redis instance for critical applications
const redis = new Redis({
  host: 'redis-server',
  port: 6379
});

// Better approach: Use Redis Cluster or Sentinel for high availability
// With Redis Cluster
const cluster = new Redis.Cluster([
  {
    host: 'redis-node-1',
    port: 6379
  },
  {
    host: 'redis-node-2',
    port: 6379
  },
  {
    host: 'redis-node-3',
    port: 6379
  }
], {
  redisOptions: {
    password: 'your-password'
  },
  // Enable read from replicas for read-heavy workloads
  scaleReads: 'all'
});

// Or with Redis Sentinel
const redis = new Redis({
  sentinels: [
    { host: 'sentinel-1', port: 26379 },
    { host: 'sentinel-2', port: 26379 },
    { host: 'sentinel-3', port: 26379 }
  ],
  name: 'mymaster', // Name of the Sentinel master set
  password: 'your-password'
});

Using a single Redis instance for critical applications can lead to single points of failure. Use Redis Cluster or Sentinel for high availability and automatic failover.

Not Using Redis for Its Intended Use Cases

// Anti-pattern: Using Redis for complex queries and relationships
// Trying to model complex relationships in Redis
redis.sadd('user:1000:friends', '1001', '1002', '1003');
redis.sadd('user:1001:friends', '1000', '1002');

// To find friends of friends (complex query)
async function getFriendsOfFriends(userId) {
  const friends = await redis.smembers(`user:${userId}:friends`);
  const friendsOfFriends = new Set();
  
  for (const friendId of friends) {
    const friendsFriends = await redis.smembers(`user:${friendId}:friends`);
    for (const fof of friendsFriends) {
      if (fof !== userId && !friends.includes(fof)) {
        friendsOfFriends.add(fof);
      }
    }
  }
  
  return Array.from(friendsOfFriends);
}

// Better approach: Use Redis for its strengths and other databases for complex queries
// Store relationship data in a graph database or relational database
// Use Redis for caching frequently accessed data

// Cache the result of complex queries
async function getFriendsOfFriends(userId) {
  // Check if result is cached
  const cachedResult = await redis.get(`cache:fof:${userId}`);
  if (cachedResult) {
    return JSON.parse(cachedResult);
  }
  
  // If not cached, query the primary database
  const result = await db.query(
    'SELECT DISTINCT f2.friend_id FROM friends f1 ' +
    'JOIN friends f2 ON f1.friend_id = f2.user_id ' +
    'WHERE f1.user_id = ? AND f2.friend_id != ? AND f2.friend_id NOT IN ' +
    '(SELECT friend_id FROM friends WHERE user_id = ?)',
    [userId, userId, userId]
  );
  
  const friendsOfFriends = result.map(row => row.friend_id);
  
  // Cache the result
  await redis.set(`cache:fof:${userId}`, JSON.stringify(friendsOfFriends), 'EX', 3600);
  
  return friendsOfFriends;
}

Using Redis for complex queries and relationships it wasn’t designed for can lead to complex application code and poor performance. Use Redis for its intended use cases (caching, session storage, rate limiting, etc.) and use other databases for complex queries and relationships.

Introduction

Getting Started

Cortex Features

Patterns

Integrations

Enterprise

Redis