Caching should reduce the cost of repeated computation or queries without hiding core business logic. Queues are appropriate for slow work such as email, image processing and service synchronization. The key is monitoring hit rate, queue latency and fallback behavior.
Redcoweb journal
Redis Cache and Queues: Where They Help and Where They Do Not
Separating cache from queues, choosing TTLs and avoiding Redis as a single point of failure.