When working with multidimensional databases, there is a problem of long time access to data. Most servers that support OLAP-technology store data on disks, thus increasing the access time to the required data. An alternative method to store data is using RAM; this, however, imposes a limitation on the amount of memory. The Palo server is organized in accordance with MOLAP principles, and uses random access memory as data storage. Access to the data and metadata of the server is currently provided only through atomic Web requests. Use of MDX extends the functionality of access to the server, and also unifies the protocol of access to data. The result of the study is the subsystem generating MDX-queries for data retrieval from multidimensional cubes of OLAP-server Palo with interactive graphical interface. The project solved the tasks of analyzing software tools for working with MDX-queries, examining the syntax of an MDX-query for data retrieval, developing algorithms for extracting data and metadata from OLAP-cubes, and implementing the MDX-query generation sub-system for retrieving data from an OLAP-cube with interactive graphical interface. The results can be used to build remote access subsystems to the Palo OLAP-server.