Перегляд за автором "Korobko, O."

Зараз показано 1 - 1 з 1
  • Зображення мініатюри
    Публікація
    Computer system efficient diagnostics with the usage of real-time expert systems
    (EWDTW, 2006) Krivoulya, G. F.; Lipchansky, A.; Korobko, O.
    The application of the modern methods of evolutional modelling for the efficient diagnostics of computer network on the basis of real-time expert systems is considered in the given article. The usage of Petri network apparatus for the purposes of expert system structure simulation is proposed. The limited Petri nets are considered to be the most effective way of simulation of processes with the binary logic. During the previous two decades the complexity and operating speed of computers have increased thousands times. The problem which demanded the year of calculations in 1983 can be solved in an hour now, and modern pocket PC is much more powerful than the desk-top one of those days. But we are to pay for the progress and computer systems (CS) improvement. The more complex computer systems become the less steady and reliable they operate. Personal computers buzz regularly, internet sites stop working rather often. New software which was designed to achieve CS productiveness frequently worsens the situation from the point of view of reliability. As a result, the annual costs on the support and the repairs of CS exceed software and hardware worth for the individuals and for the corporations as well. The group of Stanford and California Universities (USA) collaborators has elaborated new approach. They assume operator mistakes and operational system faults to be a necessity in the process of complex system functioning. Instead of the attempts to get rid of the defects the developers concentrated on the development of the system being able to restore quickly after emergency. Such an approach was called “recovery oriented computing” (ROC). By the example of the functioning of the number of Internet sites the possible reasons of their defects were investigated. The mistakes of the operators appeared to be the first reason of the system halt. Traditional measures of software and hardware reliability improvement are figured on the operators’ correct actions. But now in often happens that the system falls out due to their negligence and the time of the system standstill is longer than the idle time caused by the other troubles with hardware and software.