The Load Balancer in a Cloud Environment

The thesis was assigned by N4S@JAMK, a project within JAMK University of Applied Sciences. N4S@JAMK is also a part of research project DIGILE. N4S adopts a real-time experimental business model based upon deep customer insight.

The aim of this thesis was to explore the operation of the load balancer in a cloud environment, find out how load balancing can be used in the cloud, and what should be taken into account while deploying load balancer. Teamboard of N4S@JAMK was used as an example of a service, for which user functions were simulated. The tests aimed to identify bottlenecks of the service, so that the service could withstand more load by fixing identified bottlenecks.

The thesis carried out measurements, which tested how many simultaneous users the service can handle, how the load balancer works. It also located bottlenecks in the service. A fix was made for the service based on the results, and their benefits were verified by measurements.

As a result, the service was made scalable, and HAProxy was added to load balance the traffic for the service. By using the load balancer and by fixing the problems the service was able to handle 720 concurrent users. The limit of the concurrent users can easily be increased by balancing load on several machines.

Jarmo Viinikanoja (JAMK): Load Balancer in Teamboard 2.0-service