Harness has implemented an event-driven architecture using Redis Streams to improve communication between microservices, achieving efficient message delivery and scalability with minimal infrastructure changes, suitable for handling thousands of messages per minute. This approach addresses the limitations of synchronous communication methods like REST endpoints, offering a solution to challenges such as service degradation and retry management. Redis Streams, an append-only data structure introduced in Redis 5.0, enables real-time messaging and persistence of messages until explicitly truncated, allowing for selective consumption and explicit message acknowledgment. Harness chose Redis Streams over other options like Apache Kafka due to its existing infrastructure setup, performance advantages as an in-memory data store, and the ability to scale without introducing unnecessary complexity. However, challenges such as monitoring and persistence need careful management, as do resilience measures to handle message processing errors. The implementation has been successfully deployed in production, highlighting Redis Streams as a lightweight alternative for event-driven architectures.