Software and Hardware for Distributed Systems and Supercomputers

Article # 15_2020

47 p.

PDF

Yuriy S. Zatuliveter, Elena A. Fishchenko
Towards computer-network architectures for the digital transformation of large-scale systems

Based on the mathematical generalization of the classical model of the universal computer by J. von Neumann, the article proposes a new approach to eliminating the causes of reproduction of heterogeneity of hardware, software, and information resources in the global computer environment (GCE). The generalized model allows seamlessly extending the property of universal programmability from internal computers’ resources to arbitrarily large networks. The hardware implementation of "heavy" system functions performed programmatically within operating systems cardinally increase cybersecurity. Analysis of trends in the development of large distributed systems has shown that existing technologies for functional integration of heterogeneous GCE resources (Grid, Cloud, peer-to-peer networks) have come close to the limits of their ability to increase the scale of such systems. Further increase in their size requires unacceptable costs to overcome extreme heterogeneity and ensure cybersecurity. The reasons for the heterogeneity of GCE are fundamental. They are hidden in the logic of the von Neumann bases of microprocessor architectures. TCP/IP network protocols legalize heterogeneity on a global scale and heuristic methods for integrating heterogeneous resources. The starting principles of GCE formation were not intended for a systemic-holistic solution to the problems of creating arbitrarily large distributed systems in GCE. The proposed model opens up opportunities for implementing a mathematically homogeneous, universal, seamless programmable, and cybersecurity algorithmic space for distributed computing in the GCE. With the elimination of heterogeneity at the level of mass applications, the cost of creating/developing the entire diversity of arbitrarily large distributed systems is cardinally reduced. (in Russian)

Key words: global computer environment, large-scale systems, digital transformation, heterogeneity, integration of network resources, tree-like structure calculus, distributed computing model, seamless programming, non-microprocessor architecture, cybersecurity, universal algorithmic space of distributed computing.

article citation

DOI