03643ntm a22005177i 4500 000491235 CZ-PrVSE 20170613134144.0 m d cr n|||||||||| 170613s2017 xr fsbm 000 0 eng d NEZPRACOVANÝ IMPORT ABA006 cze ABA006 ABA006 rda Sýkora, Jan ISIS:102234 dis Analyzujte možnosti aplikace strojového učení na load-balancing eng An Analysis of Potential Applica-tions of Machine Learning in HTTP Load Balancing / Jan Sýkora 2017 ?? stran : digital, PDF soubor Vedoucí práce: Rudolf Pecinovský Diplomová práce (Ing.)—Vysoká škola ekonomická v Praze. Fakulta informatiky a statistiky, 2017 Obsahuje bibliografii Textový (vysokoškolská kvalifikační práce) Rok obhajoby 2017 Both machine learning and HTTP load balancing are well known and widely researched concepts and methods. My diploma thesis addresses possible applications of machine learning to HTTP load balancing. The main objective is to find a method to achieve better utilization of load balancing. This objective can reduce monetary costs and provide better stability of a load balanced system. In the first part, machine learning workflow and methods are described in order to analyze whether such methods could be applied to load balancing systems. After that, the current state of HTTP load balancing methods and strategies is outlined. Finally, a load balancing method using machine learning is designed and tested. The method is based on the least loaded approach using predicted values to balance HTTP traffic, the machine learning models were selected by using a grid search to find the most accurate models. These meth-ods were tested and performed well in comparison to other methods. The tests were conducted with over a hundred machine learning models, not all models were accurate or had short enough learning times. Lacking those factors deemed them unsuitable for later tests. The models were compared based on measured utilization and performance metrics for regression based machine learning models. The designed method could be applied to real world systems, however, it would require defining a domain specific metric. The applications should also employ a grid search in order to find the most accurate machine learning model. Způsob přístupu: Internet informační systémy a technologie [obor dipl. práce] diplomové práce fd132022 czenas master's theses eczenas machine learning load balancing HTTP protocol Pecinovský, Rudolf, 1954- xx0017259 ths Šlajchrt, Zbyněk opn Vysoká škola ekonomická v Praze. Fakulta informatiky a statistiky kn20010709399 dgg https://insis.vse.cz/zp/58053/podrobnosti VŠKP v InSIS https://insis.vse.cz/zp/58053 Hlavní práce https://insis.vse.cz/zp/58053/posudek/vedouci Hodnocení vedoucího https://insis.vse.cz/zp/58053/posudek/oponent/51421 Oponentura https://insis.vse.cz/zp/58053/priloha/15061 Přiloha k práci https://insis.vse.cz/zp/58053/priloha/15062 Přiloha k práci https://insis.vse.cz/zp/58053/podrobnosti dc:identifier NEPOSILAT VSKP vse58053 170601 58053