Year of Graduation
Estimation of the Data Storage System Internal State by Evaluating System Observables
The result of the work is the development of deterministic and stochastic imitation discrete-event models of the work of storage systems and their components. Based on these models, an algorithm for diagnosing failures in the work of storage was proposed and described, based on a complex combination of time series analysis approaches, nonlinear filtering, selection trees, correlation and statistical analysis. Configuring the parameters of such algorithms for real data storage is a laborious and long process. To accelerate and simplify this process, the development of a bench for testing the software complex for simulating the operation of storage systems and algorithms was started, which will allow to adjust the parameters of the algorithms for the available statistical data in an automatic mode.