In recent years, the need for an efficient and scalable framework to handle the increasing demand for energy consumption data from residential consumers in the context of smart grids (SG) has been growing. In addition, this data is increasingly massive and in large volumes, making analysis slow and decision making by the power utility impractical. In addition, there is a need to transform all historical data into profiles that determine the energy consumption pattern of consumers independently, better known as load profiling (LP). Based on this, this paper proposes a comparative analysis between microservice architecture and monolithic architecture to provide LP services in SG. As a result, the microservice architecture presented significant advantages over the monolithic architecture, standing out for the modularity, independence, and scalability offered. The adoption of microservices was recommended by this work because it allows better maintainability, evolution, and flexibility in the implementation of LP services in SGs, meeting the growing demands of data and promoting greater agility in the development of these services, besides presenting a lower cost of deployment and maintenance. However, this work contributed to decision making in the scope of system architecture for the SG scenario.