Building data centers: information and cognitive aspects

The purpose of the research is to systematize the properties of data centers and to find ways to increase the productivity of a network of centers based on collective access, combining the steps of storing and processing information and its replication. Methods . The research methods are based on the theory of organization of complex information and analytical systems and artificial intelligence systems. The application of the principles of distributed-territorial organization of centers, deferred data collection, data processing based on computational and analytical models and knowledge representation models allows you to create data center architectures reconfigurable for basic computing processes. To describe the architecture of the upper level, standard and modified information and control cycles of information processing were created and analyzed. Based on graph theory methods, the complexity of the considered cycles is estimated. At the same time, both standard quantitative and structural indicators were used, which made it possible to obtain recommendations on the organization of data centers. Results . The modified information and control cycle of network information processing is characterized by increased connectivity of vertices and internal processing cycles. A comparison of the calculated average output indicators for standard and modified cycles showed the expediency of combining steps in cycles that simulate certain human cognitive capabilities when processing a large amount of unstructured data (decomposition, aggregation, analysis, generalization). The constructed modified cycle has higher average and relative complexity indicators. Conclusion . The combination of information storage and processing steps in the information management cycle and the identification of new patterns makes it possible to interpret modern data centers as intelligent network storagesearch engines. The created modified information and control cycle allows iteratively performing the steps of systematization, generalization, and data analysis to obtain new pieces of information. This feature provides an increase in the overall productivity of the Data Center and a reduction in the cost of ownership of information resources.