Performance Evaluation of Server Cluster Load balancing Algorithms Using SDN Open Flow Model
No Thumbnail Available
Date
2018-10
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
AAU
Abstract
Due to the growth in the network usage, there is huge traffic pressure to server
clusters in the datacenters. To manage incoming traffic, operators choose a load
balancing technology to assign the incoming traffic to server clusters. However, this
classical load balancing technology has complexity and challenge in scalability,
flexibility and manageability. But, the emerging software defined networking
having application, control and infrastructure layers offers an advantage of low
cost, scalable, programmable and easy management to server clusters load
balancing activity by abstracting the low level functionality of the network.
This study evaluates the performance of round robin, weighted round robin
and least load server cluster load balancing algorithms using software defined
networking open Flow model by providing ethio telecom’s enterprise shop
customers relation Management system users’ data as input in order to answer a
research question of which load balancing algorithms is better in terms of
network performance parameters such as response time, transaction per second,
throughput in SDN platform.
Simulation has been used as a methodology to evaluate server load balancing
algorithms using Open Flow model by creating a virtual environment with oracle
virtual box. In addition, mininet simulation tool has used to create the network
topology and POX controller used in the control layer of the open Flow model
to do the performance evaluation of the load balancing algorithms. The simulation
result shows that round robin algorithm is better than weighted round robin and
least load server load balancing algorithms interms of response time (sec),
transaction rate (trans/sec), throughput (MB/sec).
Description
Keywords
Datacenter, server cluster algorithms, Mininet, SDN