Kozachuk A. Information technology of cloud application scaling that is based on forecasting of its state

Українська версія

Thesis for the degree of Candidate of Sciences (CSc)

State registration number

0416U005738

Applicant for

Specialization

  • 05.13.06 - Інформаційні технології

01-12-2016

Specialized Academic Board

Д 05.052.01

Vinnytsia national technical university

Essay

The object of study - the process of scaling cloud application. The study aims to improve the efficiency of cloud application through the use of information technology which is based on a combination of reactive and predictive approaches to horizontal and vertical scaling. Methods, used in the research include system analysis and synthesis, ARIMA, exponential smoothing, queuing systems theory, Page-Hinckley method, predictive and reactive methods of scaling, statistical methods. Theoretical results: proposed information model of cloud application scaling, which includes cloud application model as a queuing system, the classification criteria of the current mode of the cloud application, graph of its possible states, which allows to evaluate current and predict future states of cloud application; improved method of making decisions about cloud application scaling that unlike existing is based on a combination of reactive and predictive approaches to improved horizontal and vertical scaling and uses information about possible cloud application load peaks, increases power set of possible decisions about cloud application scaling; first developed information technology of cloud application scaling, which provides analysis of information about the status of a cloud application and implements an improved method of making decisions about its scaling, thus improving the efficiency of the cloud application. Practical results: algorithm of prediction of cloud application network request execution time; algorithm that determines a local maximum of cloud application network request time series; the information system of automated cloud application scaling that includes modules of collection of statistical information, forecasting the number of network requests, scaling decision making and interface to the cloud-based platform; simulator of the cloud application. Degree of implementation: the results of the research introduced in LLC "CDM Ukraine" in the form of an information system of automatic scaling of a "BrainTank" brainstormin application (№ 47/4 from 28.10.2015). Scope (industry) application - cloud computing.

Files

Similar theses