Časovna zahtevnost klasifikacije pri uporabi virtualizacije strojne opreme

(magistrsko delo)

Marko Uršič (Author), Gregor Štiglic (Mentor)

Abstract

Virtualizacija nam omogoča, da hkrati uporabljamo več operacijskih sistemov na enem računalniku, posledično nam to omogoči uporabo programske opreme, ki teče na določenem operacijskem sistemu (npr. Linux), znotraj drugega, t. i. gostiteljskega operacijskega sistema (npr. Windows). Virtualizacija je zelo primerna za testiranje in eksperimentiranje z novo programsko opremo, saj ima možnost izdelave posnetka stanja, kar pomeni, da lahko ob nastanku težav znova zaženemo virtualni računalnik tam, kjer smo posneli stanje pred nastankom težav. V magistrski nalogi smo s pomočjo virtualnega okolja in izbiro primernega operacijskega sistema poiskali najugodnejšo kombinacijo, ki omogoča čim nižjo časovno zahtevnost pri izvedbi algoritmov za klasifikacijo s področja analize mikromrež. S programom Oracle WM VirtualBox smo na strojni opremi ustvarili tri virtualne računalnike in s pomočjo programskega jezika R merili čase izvajanja algoritmov na strojni opremi in na virtualnih računalnikih z različnimi operacijskimi sistemi. Pokazali smo, da uporaba virtualizacije bistveno vpliva na časovno zahtevnost klasifikacije. Izkazalo se je, da je najmanj časa za izvajanje algoritma potreboval operacijski sistem Windows 7 Enterprise, nameščen na strojni opremi, kar je znašalo kar 31 % manj časa kot najpočasnejši operacijski sistem Linux Ubuntu, nameščen na virtualnem računalniku. Virtualizacija ima velike prednosti, vendar ko izvajamo zahtevne algoritme na velikih podatkovnih bazah, ki se izvajajo dlje časa, je virtualizacija le delno koristna, saj porabimo več časa. Zato je bolje, da takšne algoritme izvajamo kar na strojni opremi in prilagajamo operacijski sistem glede na potrebe programske opreme.

Keywords

virtualizacija;operacijski sistem;računalništvo v oblaku;

Data

Language:	Slovenian
Year of publishing:	2014
Typology:	2.09 - Master's Thesis
Organization:	UM FZV - Faculty of Health Sciences
Publisher:	[M. Uršič]
UDC:	004.946
COBISS:	2063268
Views:	1911
Downloads:	165
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	Measuring classification time complexity by using virtualization
Secondary abstract:	Virtualization allows us to simultaneously use several operating systems on one computer, and consequently allows us to use the software that runs on a particular operating system (e.g. Linux), within another, so called hosting operating system (e.g. Windows). Virtualization is very suitable for testing and experimenting with new software because it has the ability to create an image, which means that if a problem occurs, we can restart the virtual machine at the recorded state prior to the appearance of the problems. With the help of a virtual environment and the selection of a suitable operating system, the best combination which enables lower time complexity in the implementation of algorithms for classification in the field of microarray analysis was found in this master thesis. With Oracle VirtualBox we created three virtual computers. Using the R programming language criteria we measured timing of the algorithm implementation in real and virtual computers with different operating systems. We have demonstrated that the use of virtualization significantly affects the time complexity of the classification. It turns out that the least amount of time for the implementation of the algorithm is needed with Windows 7 Enterprise installed on the hardware, which amounted to as much as 31% less time than the slowest operating system Linux Ubuntu installed on a virtual computer. Virtualization has great advantages, but when running complex algorithms on large databases, which are carried out for a longer period of time, virtualization is only partially helpful because it is more time consuming. Therefore, it is preferable that one runs such algorithms directly and adapt the operating system according to the software requirements.
Secondary keywords:	virtualization;operating system;cloud computing;
URN:	URN:SI:UM:
Type (COBISS):	Master's thesis/paper
Thesis comment:	Univ. v Mariboru, Fak. za zdravstvene vede
Pages:	IX, 53 str.
ID:	8700684