Open AccessPERFORMANCE EVALUATION OF COMPUTING SYSTEM AND NETWORK OBJECTS AT DIFFERENT LEVELS OF SURVIVABILITY
Author(s) -
Andrey V. Vertemyagin,
Vasiliy G. Litvin,
Yuriy V. Litvin
Publication year - 2021
Publication title -
avtomatizaciâ processov upravleniâ
Language(s) - English
Resource type - Journals
ISSN - 1991-2927
DOI - 10.35752/1991-2927-2021-2-64-77-85
Subject(s) - survivability , computer science , queueing theory , queue , distributed computing , exponential distribution , process (computing) , service (business) , point process , poisson distribution , real time computing , markov chain , object (grammar) , markov process , point (geometry) , queue management system , computer network , artificial intelligence , statistics , geometry , mathematics , economy , machine learning , economics , operating system
The article discusses an approach to calculating the performance of objects exposed to adverse effects (AE), based on Markov models of queuing theory operating in a random environment. Three types of models of service queuing systems with an unlimited and a limited queue are investigated differing from each other in the message processing modes performed in the recovery process after AE: (1) incoming messages continue to be executed with a different performance compared to the normal mode; (2) in the process of recovery, incoming messages are accumulated, and processing is blocked; (3) in the process of restoration, no messages are received and no processing of messages already in the queue is carried out. The modes of operation of objects change as a result of random influences of the external environment (cyberattacks, influences of the external environment or object failures). The models are considered under the simplest assumptions about the distributions of the input flows of messages arriving at the object, their service time, the distributions of the time of occurrence of the AE and the recovery time of the object after that (Poisson flows and exponential time of the considered time intervals). The developed models make it possible to evaluate the design decisions taken from the performance point of view on the organization of message service for a given survivability under stationary assumptions. Numerical examples illustrate the use of models for assessing performance indicators and some indicators of survivability of the objects under study.