Programming of unloaded redundancy mode in computer systems of critical application

Today, the quality control of intermittent failures, the share of which in the total failure rate is approximately 80-90 %, as well as the correct identification of the state of these channels based on the analysis of error signals for the minimum time of their disconnection from the solution of "useful" tasks are of particular importance today in improving the efficiency of using channels of computer systems of critical application. The purpose of the study is to develop models and algorithms for analyzing the sequences of errors that result from channel failures of computer systems for critical use with majority redundancy, which will minimize the likelihood of inaccurate identification of their technical condition and are the basis of appropriate software. In the paper for analyzing the algorithms and mathematical model a duplicated system with an additional channel, which is situated in the unloaded reserve, turns on after the fault with one of the channels and begins to lose reliability at the fault-time with one of the working channels, has been considered. Knowing the intensities of failures and recoveries of the system's individual elements, a transition graph is constructed in which the vertices will be the states of the system, the edges will be the transitions with intensities of failures and recoveries. As a result of analyzing various DCH models the areas of expedient application depending on various conditions have been determined. When using low-reliability channels, an advantage of DCH₁ with a short-time system operation, taking into account failures, is observed compared to DCH₂ without taking into account failures of the system channels. For more reliable channels, it is advisable to use DCH₂ without taking into account channel failures. Analytical expressions for the structures with a "cold" backup channel are considered in more depth due to the fact that this is the most acceptable means of increasing the CSCA reliability without maintenance for a long operation's period. The reliability of computer systems of critical application, which use the mode of unloaded redundancy, is researched in terms of resistance to short-term failures, and the areas of their appropriate use are determined. The analytical expressions obtained in the article allow to make a choice of unloaded redundant structures of CSCA

software; computer systems of critical application; failure; unloaded redundancy