Architectures based in microservices have emerged as a successful style to develop software. They are particularly suitable to the cloud infrastructures, which provides several benefits, such as flexibility, modularity, resiliency, scalability, independency development and ease of deployment. However, reach the best values in each one of them, imposes several challenges. The most common include: balancing speed and safety, data synchronization, versioning, discovery, debugging, logging, monitoring, testing and fault tolerance. The problems associated with these challenges have been subject to discussions in both formal and informal literature. This talk is focused on the circuit breaker pattern, which describes a good solution to reoccurring problems fault tolerance. Although the idea behind the pattern is easy, its configuration is different for each application case and it could become complex according to the organization requirements. This talk proposes that this configuration can be addressed as an optimization problem and in this sense, a genetic algorithm is developed to define the properties required in its application.