Modeling and Research of Processing Big Data Sets in Distributed Information Systems

In this paper the features of the method of singular decomposition of large arrays of information for distributed systems are investigated. This method allows you to reduce the amount of big data by throwing away excess data. The work of the algorithm of singular decomposition in distributed systems using Hadoop and Spark technologies is simulated. As a result of a system performance study, results were obtained that show the feasibility of using Spark technology, since the data is written during processing to RAM rather than to a disk that speeds up calculations.