Redis as More Than a Cache: Queues, Sessions, and Pub/Sub

Redis: A Swiss Army Knife

Redis is an in-memory data store with sub-millisecond latency. But it is far more than a key-value cache, its built-in data structures make it a versatile tool.

Queues with Redis

Laravel Horizon uses Redis lists and sorted sets for job queues. The advantages over database queues:

• Performance: Redis processes millions of jobs per second, no database locks, no polling.
• Priorities: Multiple queues with configurable worker assignments.
• Monitoring: Horizon dashboard with real-time insight into queue utilization and errors.

Sessions and Rate Limiting

• Sessions: Redis sessions are faster than database sessions and survive deployments (unlike file sessions).
• Rate Limiting: Redis INCR with TTL is the foundation for Laravel's rate limiter, atomic and fast.
• Distributed Locks: Cache::lock() uses Redis for mutex operations across multiple servers.

Pub/Sub and Streams

• Pub/Sub: Real-time messaging between services, ideal for Laravel Broadcasting with WebSockets.
• Streams: Persistent message queues with consumer groups, similar to Kafka but easier to operate.

Our Recommendation

In every Laravel project, we deploy Redis, at minimum for cache and sessions, often also for queues and broadcasting. With Redis Sentinel or Cluster, high availability is not an issue either.