In the past decade there is a significant trend of implementing IoT technologies and standards in different industries. This trend brings cost reductions to the companies and other benefits as well. One of the main benefits is real-time and uniform data collection. The data are transferred using diverse communication protocols, from the sensor nodes to the centralized application. So far, current approaches in developing applications are not proved itself to be efficient enough in scenarios when a significant amount of data needs to be stored and analyzed. The focus of this paper is on development of software architecture suitable for usage in Internet of Things (IoT) systems where the larger amount of data can be processed in real-time. The software architecture is developed in order to support the sensor network for monitoring the small data center and it is based on microservices. Besides the system and its architecture, this paper presents the method of analysis of system performances in real-time environment. The proposal for lightweight microservice architecture, presented in this paper, is developed with .NET Core and RabbitMQ, with the utilization of MongoDB and SQLite databases systems for storing data collected with IoT devices. In this paper, the system evaluation and research results in different stress scenarios are also presented. Because of its complexity, only the most significant segments of architecture will be presented in this paper. The proposed solution showed that proposed lightweight architecture based on microservices could deal with the larger amount of sensor data in the case of using MongoDB. On the other hand, the usage of SQLite database is not recommended due to the lower performances and test results.