Relational and noSQL storages are developed for the fast processing of the large data sets having a stable structure, while the ontologies are used to rep-resent complex and dynamic sets of information of a limited size. In the in-dustrial applications it is often needed to maintain the large warehouses of data consolidated from various sources. The ontologies are useful to repre-sent the structure of that data, but RDF triple stores are not well suitable for storing it. We offer an approach and a system allowing to use the opportuni-ties of fast storage engines along with the flexibility of ontology-based data management tools, including SPARQL queries. The system implements a multi-model data abstraction layer which allows working with the data as if it is situated in RDF triple store, executes SPARQL queries over it and ap-plies SHACL constraints and rules.