T.V. Afanasieva  Dr.Sc. (Eng.), Professor of Department of Information systems, Ulyanovsk State Technical University. E-mail: tv.afanasjeva@gmail.com
A.O. Kozlov  Post-graduate Student, Ulyanovsk State Technical University. E-mail: alexander.o.kozlov@gmail.com

Now at the enterprises types of IT resources are used various by the nature of solved tasks, decrease or which loss of working capacity has key value for business processes. One of frequent problems which IT services of the enterprises face is a failure of the server equipment. Technical malfunctions, «falling», blackout, human factor, incorrect parameters established software, increase in loading at the server, gradual changes of the physical characteristics leading subsequently to refusals can be the reasons. The carried-out analysis shows that existing forecasting methods of server parameters operate only with accurate (numerical) data that means which results can be interpreted by various experts differently. In the present article the integration of mathematical time series analysis methods and fuzzy expert estimates for forecasting of the server behavior is considered. The algorithm of regression and fuzzy modeling for forecasting of the server behavior is offered. Received results are presented in a linguistic form, based on mathematical models of time series and fuzzy tendencies and therefore are more reliable in comparison with expert estimates of administrators of IT resources. On the other hand, in comparison with forecasting techniques on the basis of mathematical methods of the analysis of numerical time series offered approach has advantage as in it the forecast is specified at the expense of external expert estimates. Practical application of the offered algorithm - detection of the situations leading to decrease in operability of IT resources at early stages and obtaining information on them, necessary for adoption of reasonable and operational administrative decisions.