Measurement based analysis of applications running in virtual environments

OData support
Supervisor:
Dr. Maliosz Markosz
Department of Telecommunications and Media Informatics

The control of the resource consumption of the virtualized applications is an important need for the service operators. Based on this need I proposed a measurement framework that is able to monitor the performance of applications running in different virtualization environments.

In the first part of my Thesis I presented the virtualization technologies, with a focus on platform and operating system level virtualization. I reviewed the well-known information system monitoring applications, selecting the most customizable one as the basis for my solution.

During my work I used an open source application (ndpireader) implementing the deep packet inspection functionality as Virtualized Network Function (VNF) with. I set up three different virtualization environments to investigate the VNF performance, selecting the three widely used systems: the OpenStack cloud environment, the Docker container virtualization system and the QEMU-KVM system. I designed an automated monitoring system that is able to measure the resource usage of a given application through the processor load, memory load and network traffic load parameters. The advantages and drawbacks of the three virtualization technologies can be assessed by comparing the results of the measurements.

As future work I plan to test other, more complex network virtualization functions, as well.

Downloads

Please sign in to download the files of this thesis.