Handling media flows in virtual environments

OData support
Supervisor:
Dr. Simon Csaba
Department of Telecommunications and Media Informatics

The goal of my Thesis is to assure the scaling of an application running in a virtual envi-ronment, based on the resource usage of the application. My solution is based on clouds, because the cloud-based technologies solve the management of resources of the virtual machines running in the cloud.

In the first part of my Thesis, I studied the relevant literature and described the charac-teristics of the virtualization solutions. I presented the OpenStack cloud system, the Docker container virtualization, and the VMware virtualization solutions in greater de-tail. I reviewed the currently available dynamic load balancing solutions available on the market. I have reviewed the widely used video on demand streaming applications and I selected VLC because it is freely available, open source, cross-platform, and supports numerous video and audio formats. I designed an OpenStack cloud system for my ex-periments that supports scalability. I solved the generation of dynamic load with the help of bash scripts and collected the resources usage reports with the help of the top application. I performed measurements of a video on demand service with RTSP and HTTP streams. During the evaluation of the experiments I observed that the increasing number of RTSP stream requests resulted in a linear increase in terms of CPU load on the servers, but above a certain level of demand the growth in the resource usage slowed down. In the case of HTTP streaming the increased number of clients resulted in a sig-nificantly lower overhead, because under this streaming alternative the media server streams the same video for each client.

My future work will focus on the analysis of the occurrent questions, and will include the development of a solution that addresses these issues.

Downloads

Please sign in to download the files of this thesis.