Porazdeljeno procesiranje, analiza in vizualizacija podatkov z mehanizmi visoke skalabilnosti

GAČNIK, MATEVŽ (author), Marolt, Matija (mentor)

V nalogi smo predstavili konceptualni in izvedbeni model za skalabilno, porazdeljeno in uteženo izvajanje velike količine operacij na več procesnih enotah, ki delujejo v oblaku. Delo prikazuje način razvoja sistemov za procesiranje, ki imajo minimalne procesne, kot tudi časovne omejitve ob zahtevi po spremembi količine procesne moči. Razloženi so načini implementacije elastičnega prilagajanja glede na potrebe z uporabo procesiranja v oblaku. V delu smo preučili načine za filtriranje uporabnih podatkov po kriterijih, ki so za problemsko domeno pomembni. Prikazane so možnosti za napredno filtriranje podatkov glede na zahteve analiz. Delo podaja tudi načine in izvedbo vizualizacije naprednih analiz nad zajetimi in procesiranimi podatki v obliki intuitivnih in interaktivnih grafov, grafikonov, besednih oblakov ali drugih za uporabnika primernih prikazih.

• Language: Slovenian
• Keywords: porazdeljeno procesiranje, analiza, paralelizem, oblak, računske operacije, skalabilnost
• Work type: Master's thesis
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2016
• PID: 20.500.12556/RUL-85017
• Publication date in RUL: 09.09.2016

Secondary language

• Language: English
• Title: Parallel data processing, analysis and visualization using high scalability mechanisms

Abstract

In this work we present conceptual and implementation model for scalable, distributed and balanced execution of large number of compute operations running on multiple processing units in the cloud. We provide system development methods for large scale processing with minimal time constraints and limitations in regard to increasing scale-out parallelism in the cloud. Implementation details regarding elastic adjustment to processing units are discussed in connection to required processing power needed in a cloud environment. Work provides filtering approaches for useful data in the described problem domain. We present options for advanced data filtering in multiple stages, which correlate with needed analyses requirements. At the end of this work we present ways of visualization of advanced analysis of gathered data in a form of intuitive and interactive UI components, graphs, word clouds and other user acceptable views.

• Keywords: parallel processing, analysis, parallelism, cloud, compute operations, scalability