Nowadays everyone uses internet that cause higher traffic on internet network. High traffic causes server overload and service not available. This problem can be solved by implementing multiple web servers with a load balancing mechanism. Load balancing is mechanism how traffic will distributed by using certain methods to optimize network. Load balancing in Software Defined Networking (SDN) handles big traffic load and complexity issues so it improves server performance. Performance of server is affected by resource usage from each server. Resource usage includes CPU, Disk and Memory usage. Information of resource usage is obtain from psutill agent. This research applies load balancing using K-Nearest Neighbor (K-NN) algorithm on POX controller to decide which server will handle incoming traffic. Traffic will distribute to server which has smallest value from CPU, Disk and Memory usage. Performance testing uses two scenarios, first is different rate number (req/s) and second is different load in server. Result from this research showed if K-NN algorithm can distribute traffic to server which has smallest status. In throughput test showed if increasing of rate (req/s) also make increasing of throughput. The highest value of throughput get in 500 req/s with value is 207.3 KB/s. Response time test showed if increasing of rate and load in web server also make increasing of response time. It happens because server receive and handle much request from client. Best response time get in this research is 3.9 ms.
Copyrights © 2020