Today, most companies use databases as a common place to store and manage different types of data. The goal of database usage is to transform acquired data into information through creation of various analysis and building predictive models, which could forecast specific outcomes and provide inputs for business decisions. Relational databases are the most commonly used across industries but today's dynamic situations have also arisen a new type of database model – non-relational (NoSQL) databases, which are becoming increasingly popular in big data industries. While relational databases are mostly convenient for structured, homogeneous data, NoSQL databases are used with large datasets that typically have a different, non-standard structure. This paper describes both relational and NoSQL databases: their types, characteristics, advantages and disadvantages, with the focus on maintaining their security. Generally, protecting NoSQL databases is more complicated than protecting relational databases due to their distribution, lack of consistency, dynamic and unstructured data. An example of NoSQL databases provided in this paper are MongoDB and Cassandra. In the case study, a MongoDB database was created for the needs of the Croatian Meteorological and Hydrological Service in order to show necessary steps for the process of designing and installing NoSQL database security measures.