EN
NoSQL databases become more and more popular, not only in typical Internet applications. They allow to store large volumes of data (so called big data), while ensuring fast retrieving and fast appending. The main disadvantage of NoSQL databases is that they do not use relational model of data and usually do not offer any declarative query language similar to SQL. This raises the question how NoSQL databases can be used for OLAP processing and other Business Intelligence tasks. In the paper the author presents the most common types of NoSQL databases, describes MapReduce paradigm and discusses models of OLAP processing for such databases. Finally some preliminary results of aggregation performance in nonrelational environment are presented.