S.A. Skachkov – Dr.Sc.(Eng.), Professor, Main Research Scientist,
JSC «Concern «Morinsys-Agat» (Moscow)
E-mail: info@concern-agat.ru
O.N. Andreeva – Dr.Sc.(Eng.), Associate Professor, Professor,
RTU MIREA (Moscow)
A.V. Klyuev – Ph.D.(Eng.), Associate Professor, Senior Research Scientist,
Vasilevsky Military Academy of Air Defense Forces RF (Smolensk)
A.A. Kovalev – Senior Research Scientist,
Vasilevsky Military Academy of Air Defense Forces RF (Smolensk)
D.D. Bakhteev – Post-graduate Student,
RTU MIREA (Moscow)
When creating large systems, the three most important target dimensions of development are usually discussed – cost, performance, and reliability. Often a problem arises, which is that the dimension of reliability of a computing system (CS) is less clear than performance, that is, it is not clear what is included in the concept of reliability and what indicators it is evaluated. Reliability in the broad sense is a complex characteristic of the object under consideration, which for the consumer is one of the main indicators of quality. Reliability assessment is carried out by a set of individual and/or generalized quantitative indicators, depending both on the characteristics of the object under consideration and on the conditions of its operation. Reliability issues are addressed in a multitude of regulatory documents in individual subject areas, in sections related to quality requirements and quality management. In these cases, the definitions apply to objects of a given subject area, and CS is considered as part of the object. You can find different general definitions of reliability that do not have fundamental differences, in particular: reliability – the property of an object to keep in time, within established limits, the values of all parameters characterizing the ability to perform the required functions in specified modes and conditions of use, maintenance, storage and transportation. As can be seen from the above arguments, reliability primarily focuses on the performance of its functions by technical hardware objects, and the concept covers reliability, durability, maintainability, and storage as basic indicators that characterize reliability and have their own dimensions, for example, mean time between failures , average life, average recovery time, average shelf life. Values of reliability indicators are usually determined using two basic approaches: the study of the physical foundations of reliability - the analysis of the physicochemical processes occurring in the product and the use of the mathematical theory of reliability - the study of statistical probabilistic patterns of failure flows.
Goal of the work is to propose a model for assessing the reliability of CS.
The problem of quantitative assessment of failure tolerance in the conditions of different compatibility and overlap of types of redundancy is solved. The effectiveness of the fault detection system is evaluated. The coefficient of functional load of the CS subsystems is determined. As a result of modeling, a variant with a specific redundancy architecture that has a minimum of objective function is selected from the selected CS implementation options.
- Zayavka na izobretenie № 2018109822 ot 20.03.2018. Sposob obespecheniya nadezhnosti vychislitelnykh sistem pri razlichnykh arkhitekturakh izbytochnosti. (In Russian).
- Mikhalevich V.S., Volkovich V.L. Vychislitelnye metody issledovaniya i proektirovaniya slozhnykh sistem. M.: Nauka. Gl. red. fiz.-mat. litry. 1982. (In Russian).
- Klyuev A.V. Dis. … k.t.n. na spetsialnuyu temu. Smolensk: VA VPVO VS RF. 2007. (In Russian).
- Skachkov S.A. Metody, modeli i sredstva povysheniya nadezhnosti ZRV voiskovoi PVO v usloviyakh otkazov sboinogo kharaktera vychislitelnykh sistem: Monografiya. Smolensk: VA VPVO VS RF. 2008. 273 s. (In Russian).