Performance Analysis High Availability of Web Server in Cluster GKE (Google Kubernetes Engine) using Infrastructure Google Cloud Platform

Abstract

Internet users in the world according to the Internet World Stat in 2020 reached 58.7% of internet users worldwide, this percentage increases from year to year. The growth of internet users worldwide, namely 1,167% from 2000 to 2020. The large number of internet users makes web service providers have to take resources for managed web servers. Based on these problems, we need a system that can help server performance and availability. Server grouping supports a number of computers to work together to perform the computation process. The technology server clustering has the advantage of being able to produce a system with a high level and level. Kubernetes is a cluster platform for container or container orchestrators. Kubernetes is expected to be a solution for more ef icient computing and the creation of a high-level system. The simulation system uses the public cloud platform, namely the google cloud platform with kubernetes services with google kubernetes engine services. The system is simulated using the google kubernetes engine on the google cloud platform. The average value of Availability was 99.96% of the total 27526 minutes of uptime. With the computational load of the nodes on the google kubernetes engine, the average CPU usage for each test scenario is quite stable with the highest CPU at 27.178% From the test result of throughput, the average value of throughput on each connection is 7,126 Mbit / s with the highest value for 1000 connections, namely 8,764 Mbit / s. The delay indicates the "Very Good" category based on the TIPHON standard of less than 150 ms. The highest percentage of packet loss was obtained at 5000 connections at 16.27%.