Message queues are asynchronous communication components that enable distributed systems to exchange data reliably without requiring direct connections between services. They act as buffers between producers and consumers, storing messages until they can be processed, which decouples system components and allows them to scale independently. Message queues solve critical problems in modern architectures: handling traffic spikes, enabling fault tolerance through retry mechanisms, and providing eventual consistency across distributed services. A key insight to keep in mind: choosing between message brokers (like RabbitMQ) and streaming platforms (like Kafka) fundamentally depends on whether you need transient messaging with acknowledgments or persistent event logs with replay capability — this distinction shapes your entire architecture.