Adaptive load balancing in virtual environment

OData support
Dr. Szatmári Zoltán
Department of Measurement and Information Systems

By the increasing popularity of the web services, more and more users need to be served, so the demand for increased performance, and reliability among these applications have also became vital. In order to keep financial costs low, we have to minimaliize the operational costs, however, this should not disadvantage Quality of Service parameters. The key for efficient operation can be the combination of virtualization technology and the distributed systems.

With the low-cost solution of Cloud, which can be formed by distributed systems, and which hides data, programs, and hardware from users, shared IT resources can be easily accessible with minimal administration effort, or provider intervention. The most common services, which can be published by Clouds, are Software (SaaS), Platform (PaaS), and Infrastructure (IaaS).

Whether it is a traditional IT service such as a database, a web server, or an e-business service, the operator has interest in a cost-effective, reliable working computer system.

During the process of load balance, we assemble a cluster of independent computers, which will be accessed through a common entry point. The requests to this address will be distributed between the nodes by a control element. By the implementation of the adaptive balancing technology we achieve better scalability of performance, since it allows us to suit the number of servers to the current resource usage.

I attempt to present solution for dynamic load balanced web application running on virtual computers. First, I present webservers, as the web application running units, then proxy server, which means that the servers will be used in load-balancing method, and finally a load measuring device: a monitoring system. I will show its features, its usage, and its place in the architecture. I will discuss elements of the architecture, background methods, and the range of applicability. I will give details of the operation process, the implementing algorithms, scalability, and will prove the correct functioing of the system by mesuring results.


Please sign in to download the files of this thesis.