Cloud computing has reshaped the ICT industry. Since it mainly offers flexible resources and reducing capital expenses, many companies are already migrating their systems from the previous traditional way into cloud computing architecture. The demand in the ICT industry for cloud computing is to have a low latency network and able to resolve a huge number of workloads without having any issues to the end user. Load balancing is the key component that making sure every workload of the system is distributed across multiple servers, so the job processing and response time is always at its optimal rate.
This research uses Kubernetes as the cloud computing platform with kube-proxy as its load balancing tool and the web servers are stored and operated in the form of pods inside Kubernetes cluster. The proposed system is simulated in order to test the throughput, elapsed time, server processing time, CPU utilization, and the comparison of its load balancing performance between two worker nodes and four worker nodes of the implemented load balancing system.
The results show that in elapsed time parameter, shortest expected delay has the fastest time, which is 216.295ms. In server processing time, shortest expected delay resulted 214.257ms, which is faster than least connection algorithm. The throughput when using shortest expected delay algorithm is bigger, averagely 560.256 packets/second. Least connection algorithm has the efficiency of 35.24% in terms of CPU usage, which is the lowest percentage. Scaling up its cluster, from two worker nodes into four worker nodes resulting in a significant decrease of server processing time, which means the load balancing using 4 worker nodes is more effective.
Keyword: cloud computing, Kubernetes, container, load balancer, web server, lbaas