Analisis Perbandingan Performansi Web Server Menggunakan Horizontal Pod Autoscaler Pada Container Runtime Docker Dan CRI-O Di Kubernetes

Zuhair, Yahya (2021) Analisis Perbandingan Performansi Web Server Menggunakan Horizontal Pod Autoscaler Pada Container Runtime Docker Dan CRI-O Di Kubernetes. Undergraduate Thesis thesis, Institut Teknologi Telkom Purwokerto.

	Text Cover.pdf Download (1MB)
	Text Abstract.pdf Download (217kB)
	Text Abstrak.pdf Download (218kB)
	Text BAB I.pdf Download (622kB)
	Text BAB II.pdf Download (942kB)
	Text BAB III.pdf Download (888kB)
	Text BAB IV.pdf Restricted to Registered users only Download (1MB) | Request a copy
	Text BAB V.pdf Download (395kB)
	Text Daftar Pustaka.pdf Download (572kB)
	Text Lampiran.pdf Restricted to Registered users only Download (385kB) | Request a copy

Abstract

The number of requests that enter the web server results in the use of cpu on the server experiencing an overload which causes the server to down. Kubernetes technology is used for the management of application workloads isolated in containers. The Horizontal Pod Autoscaler feature is the answer to adding and removing pods automatically. However, by default kubernetes uses the docker runtime container but kubernetes doesn't just use docker as the runtime container. There are several runtime containers in Kubernetes such as Docker runtime containers, CRI-O, Rkt, Containerd, and Frakti. Based on the architecture, the docker runtime container is more complex than the shorter cri-o container runtime because of the Container Runtime Interface via cri-o then runc. In this study, we discuss Horizontal Pod Autoscaler on the docker and cri-o runtime containers in terms of performance. The test was carried out with a test scenario of 15 times with requests of 1000, 2000, 3000, 4000 and 5000 and the number of requests per second of 100, 200, 300 400 and 500. The results showed the use of the docker runtime container cpu was 13.60%, while the critical runtime container o 8.20%. The throughput of the cri-o runtime container is 263.3 KB/s while the docker runtime container is 260.6 KB/s. The results of scaling up container runtime docker 18.2s (18.2 seconds) and scaling down 4m 75s (4 minutes 75 seconds) while scaling up container runtime cri-o 12.8s (12.8 seconds) and scaling down 4m 21s (4 minutes 21 seconds). The test results show that the cri-o runtime container is superior to the docker runtime container in terms of cpu usage performance, throughput, scaling up time and scaling down. Keywords: Kubernetes, Horizontal Pod Autoscaler (HPA), Container Runtime, Pod, Docker, CRI-O

Actions (login required)

View Item