Integrasi Network Booting dan Docker dalam Custom Linux untuk Resource Provisioning

Abdillah, Fachrendy Zulfikar (2026) Integrasi Network Booting dan Docker dalam Custom Linux untuk Resource Provisioning. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5025201018-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Penelitian ini merancang dan mengimplementasikan sistem resource provisioning otomatis untuk komputer-komputer idle di laboratorium Departemen Informatika ITS melalui integrasi network booting berbasis iPXE, custom Linux, dan Docker. Sistem menggunakan server pusat yang menyediakan layanan DHCP (Kea DHCP), TFTP (tftpd-hpa), dan HTTP (Apache2) untuk memungkinkan komputer klien melakukan booting tanpa media penyimpanan lokal. Custom Linux berbasis Debian 12 (Bookworm) dibangun menggunakan live-build dengan mekanisme hook chroot yang mengintegrasikan Docker Engine secara bawaan, sehingga klien yang berhasil melakukan booting langsung memiliki lingkungan kontainerisasi siap pakai tanpa instalasi tambahan. Proses booting klien diawali dengan perolehan alamat IP melalui DHCP, pengunduhan skrip boot.ipxe via TFTP, kemudian pengunduhan kernel (vmlinuz), initial ramdisk (initrd.img), dan sistem berkas live (filesystem.squashfs) melalui HTTP, yang dimuat dengan mekanisme OverlayFS untuk operasi diskless yang non-invasif terhadap sistem lokal yang sudah terpasang. Evaluasi performa jaringan dilakukan menggunakan Wireshark pada lima skenario pengujian dengan jumlah klien bervariasi dari 1 hingga 5 yang melakukan booting secara bersamaan. Hasil pengukuran menunjukkan utilisasi jaringan meningkat secara proporsional dari 5,10% pada 1 klien menjadi 18,40% pada 5 klien, dengan throughput rata-rata meningkat linear dari 51 Mbps menjadi 184 Mbps, mengindikasikan server HTTP mampu mendistribusikan bandwidth secara paralel tanpa saturasi. Total data yang ditransfer meningkat linear sesuai jumlah klien, mencapai 4,50 GB pada skenario 5 klien. Durasi booting rata-rata berkisar antara 81,9 hingga 124,9 detik dengan variasi signifikan antar klien yang konsisten di setiap skenario, mengindikasikan bottleneck utama sistem terletak pada sisi klien (perbedaan kondisi hardware dan infrastruktur) bukan pada server atau jaringan. Sistem berhasil memenuhi seluruh tujuan perancangan: network booting otomatis tanpa media penyimpanan lokal, integrasi Docker bawaan, monitoring, serta evaluasi performa yang menunjukkan server mampu melayani hingga 5 klien simultan dengan utilisasi jaringan masih jauh di bawah kapasitas maksimum 1 Gbps.
=====================================================================================================================================
This research designs and implements an automated resource provisioning system for idle laboratory computers at the Department of Informatics ITS through the integration of iPXE-based network booting, custom Linux, and Docker. The system employs a central server providing DHCP (Kea DHCP), TFTP (tftpd-hpa), and HTTP (Apache2) services to enable client computers to boot without local storage. A custom Debian 12 (Bookworm) Linux distribution was built using live-build with chroot hooks to integrate Docker Engine natively, allowing booted clients to possess a ready-to-use containerization environment without additional installation. The client booting process begins with IP acquisition via DHCP, download of the boot.ipxe script over TFTP, followed by retrieval of the kernel (vmlinuz), initial ramdisk (initrd.img), and live filesystem (filesystem.squashfs) via HTTP, loaded with OverlayFS for non-invasive diskless operation. Network performance evaluation utilized Wireshark across five test scenarios with one to five simultaneously booting clients. Measurements reveal network utilization increases proportionally from 5.10% with one client to 18.40% with five clients, while average throughput linearly rises from 51 Mbps to 184 Mbps, indicating the HTTP server distributes bandwidth in parallel without saturation. Total data transferred scales linearly with client count, reaching 4.50 GB in the five-client scenario. Average boot durations range from 81.9 to 124.9 seconds; however, significant and consistent per-client boot time variance across all scenarios indicates the primary bottleneck resides on the client side specifically, differing hardware conditions and infrastructure rather than the server or network. The system successfully meets all design objectives: automated network booting without local media, native Docker integration, and performance validation demonstrating reliable service for up to five simultaneous clients with network utilization well below the 1 Gbps capacity limit.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Network Booting, iPXE, Docker, Custom Linux, Debian, Resource Provisioning, Wireshark, OverlayFS
Subjects:	T Technology > T Technology (General) > T58.5 Information technology. IT--Auditing T Technology > T Technology (General) > T58.6 Management information systems
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Fachrendy Zulfikar Abdillah
Date Deposited:	04 Aug 2026 02:37
Last Modified:	04 Aug 2026 02:37
URI:	http://repository.its.ac.id/id/eprint/142703

Actions (login required)

View Item