This paper contains ideas on how to optimize the costs of running a microservice system. Currently, there is much done to provide high fault tolerance of a microservice and a system as a whole. Cloud providers come up with new ways to guarantee the high speed of newly launched instances. This leads to a ubiquitous run of redundant servers with possible cold or hot standby mode. This is often crucial because the ability to use some applications quickly and on time can be important to many users, potentially saving lives. At the same time, it's important to prioritize ecological preservation and minimize overuse of the Earth's resources. In the context of cloud, and specifically, server computing, that would involve using resources in a way that extends their lifespan, minimizing the creation of slowly decomposing waste, and avoiding excessive energy consumption. Cloud providers, such as Amazon, Google, and Azure, discard millions of underused hardware units due to the necessity of ensuring service guarantees to their customers. In the article, method to optimize the usage of servers by organizing microservices in complementary sets are described. As a result, server resources will be used most efficiently. The method of grouping the microservices can be likened to the principles of lattice theory. The ideas in the article could be useful for the systems like Kubernetes scheduler in the stage of picking the right set of instances to run a new microservice, or to cloud providers. As a result, less energy and hardware resources will be used to provide the same quality of fault tolerance.