Context: Microservice architectures offer advantages such as scalability and independent deployment compared to traditional monolithic systems. Nonetheless, the specific impact of asynchronous communication patterns within microservice architectures on critical quality attributes, such as resilience and robustness, particularly when directly compared to monolithic systems under failure conditions, remains insufficiently explored empirically. Objective: This study empirically examines the influence of asynchronous communication mechanisms on the resilience and robustness of microservice architectures in comparison to equivalent monolithic architectures, identifying associated benefits and operational trade-offs. Method: Two equivalent systems representing a simplified e-commerce order process were developed: a synchronous monolith and an event-driven microservice architecture utilizing asynchronous communication. Controlled Chaos Engineering experiments simulating service crashes were conducted under load. Key metrics, including failure rate, system availability, request duration, and processed requests, were measured and compared across architectures under identical fault scenarios. Results: The asynchronous microservice architecture demonstrated significantly enhanced resilience and robustness under simulated failures. Compared to the monolith, it exhibited lower failure rates (up to a 12% reduction observed), maintained higher availability, and processed requests more effectively during fault conditions. Asynchronous decoupling localized failures and facilitated continued partial operation. However, the microservice architecture exhibited increased operational complexity and higher resource requirements. Conclusion: Asynchronous communication substantially improves the resilience and robustness of microservice architectures against failures compared to traditional monolithic approaches. While offering significant fault tolerance advantages, the adoption necessitates careful consideration of increased operational complexity, resource consumption, and the need for sophisticated monitoring and debugging infrastructures.