Performance Analysis of a Software-Based Switch in Cloud Environment

OData support
Supervisor:
Kökényesi Tamás
Department of Automation and Applied Informatics

In the 21st century, virtualization is one of the most prominent trends in the world of IoT. Previously, dedicated hardware has been allocated to each task. Nowadays, the main goal is that we solve most of the tasks by software. There are many benefits of this new paradigm. For example, the increase of the hardware utilization, more scalability and simplified maintenance.

The network devices are no exceptions. In cloud computing technology, virtual switches are one of the most significant components. These applications provide the external and internal connections for the virtual machines running on servers.

The virtualization of physical devices adds stochastic parameters to the system, which have impact on the network. If we would like to ensure the predictability, reliabilty and other quality characteristics, we have to perform complex measurements. The purpose of the measurements are to find connection among network and environment variables, well the performance characteristics.

In my paper I present the types and behaviour of the open source software based switch implementations. I study the network traffic generator applications which are necessary for performance measurements. I find out about the measurable parameters able to represent the switch’s performance. I work out a suitable, comparable performance measurement method, then I perform the measurements with a chosen switch implementation and a traffic generator, in different setups. After the tests I evaluate the results, and map the connections among the switch’s parameters and the environment variables. In the end of my thesis, I summarize the results, and examine the automation options, regarding the testing and the evaluation.

Downloads

Please sign in to download the files of this thesis.