The development of monolithic systems brings several challenges related to system maintenance and scalability. To mitigate this problem, there is currently a trend in developing systems to use the composition of microservices. The application is divided into independently deployable services, which can quickly implement on any infrastructure resource, and each service runs in an isolated process. Furthermore, the development of technology for microservices also increases the operational complexity associated with modern applications. The service mesh is a promising approach to mitigate this situation, introducing a dedicated infrastructure layer over microservices without imposing modifications to the service implementations. The proposed work aims to present a survey about service mesh platforms and application benchmarks based on microservices and performance tests performed. As a result, tests show that despite the benefits of the service mesh, it impacts application latency.