Please use this identifier to cite or link to this item:
Title: Experimental study of data center network load balancing mechanisms
Other Titles: Πειραματική μελέτη μηχανισμών εξισορρόπησης φορτίου σε κέντρα δεδομένων
Authors: Κωνσταντινίδης, Μενέλαος
Keywords: Load Balancing
Data Centers
Software-Defined Networks
Issue Date: 2021
Publisher: Πανεπιστήμιο Μακεδονίας
Abstract: Load balancing on computer networks is a technique used to spread data traffic or task load across multiple network links or servers. In data centers the load balancing of the traffic is crucial for their efficient and smooth operation. This thesis aims to compare two load balancing algorithms, a static and a dynamic in order to find out which performs better. The background of the thesis includes the theoretical study of data centers. Their network, topologies and most important architectures are explored. Also, load balancing is included, with a focus on its application on data centers and the classification of load balancing algorithms. Moreover, some load balancing algorithms are presented with a focus on their procedure and logic. In the experiment a 3-level fat tree topology was simulated using the Mininet simulator. This network was connected to the Floodlight controller, in which the two load balancing algorithms used in the experiment were implemented along with the monitoring. The experiment included scenarios that stressed the system with the help of the D-ITG traffic generator in order to generate results from the monitored metrics. Based on these results, the efficiency of the two algorithms is evaluated and analyzed.
Description: Πτυχιακή εργασία--Πανεπιστήμιο Μακεδονίας, Θεσσαλονίκη, 2021.
Η βιβλιοθήκη διαθέτει αντίτυπο της πτυχιακής μόνο σε ηλεκτρονική μορφή.
Rights: Attribution-NonCommercial-NoDerivatives 4.0 Διεθνές
Appears in Collections:Τμήμα Εφαρμοσμένης Πληροφορικής (Π)

Files in This Item:
File Description SizeFormat 
KonstantinidisMenelaosPe2021.pdf2.36 MBAdobe PDFView/Open
KonstantinidisMenelaosPe2021presentation.pdf2.28 MBAdobe PDFView/Open

This item is licensed under a Creative Commons License Creative Commons