The growth in the volume of unstructured data generated by various applications and services, and the acceptance by companies of the fact that such data is valuable, have led to the demand for systems that can analyze large amounts of data without human intervention. Data mining systems can satisfy this need, but improving the efficiency of such systems is an urgent task. Despite the growing popularity of NoSQL solutions, the main database management systems are still relational databases. In the article, special attention is paid to the fact that modern RDBMS can be used not only as reliable data stores. Currently, the primary task of RDBMS development is to integrate data mining into them. Due to the fact that the data remains in the storage, the system does not waste resources on unloading the analyzed data set from the database and loading the analysis results back. This approach will increase both the speed of development, due to the use of services embedded in the DBMS, and the performance of the entire system. The article discusses the main problems of data mining and the existing algorithms for solving them. The main methods of implementing data mining in a DBMS are described. Special attention is paid to the approach in which the data analysis system is considered as an internal DBMS service. The paper presents well-known data analysis systems and libraries developed for RDBMS, as well as variants of SQL query language extensions.