Sari, Friska Anggia Nurlita (2018) Analisis Perbandingan Kinerja Algoritme Fair Share Scheduling Dengan Capacity Scheduling Terhadap Pengiriman Job Pada Hadoop Cluster Multinode. Sarjana thesis, Universitas Brawijaya.
Abstract
Hadoop merupakan sebuah framework berbasis open-source digunakan untuk pengolahan data dengan skala besar dan tersimpan dalam sekelompok komputer yang saling terhubung dalam suatu jaringan secara terdistribusi. Hadoop terdapat beberapa algoritme penjadwalan untuk mengatur antrian job yang masuk, job yang telah di-submit oleh pengguna akan saling berkompetisi untuk merebutkan suatu resource yang tersedia. Sehingga perlu adanya algoritma penjadwalan yang bertugas mengatur jalannya job pada sebuah resource serta mengelola pemrosesan data agar hasil yang dikeluarkan sesuai dengan apa yang diharapkan. Penelitian ini dilakukan untuk mengetahui perbandingan dari kinerja algoritme Fair Share Scheduling dan Capacity Scheduling pada pengiriman job pada Hadoop multinode untuk mengetahui nilai dari parameter job fail rate, latency, dan throughput. Capacity Scheduling mendukung antrian secara hirarki yaitu dapat membagi resource yang tersedia pada cluster ke beberapa antrian. Fair Share Scheduling adalah metode alokasi pembagian resource pada antrian secara adil untuk seluruh job yang masuk pada antrian. Berdasarkan hasil pengujian yang telah dilakukan, memiliki nilai minimal Job Fail rate yang lebih baik dibandingkan Fair Share Scheduling sebesar 0,99%. Fair Share Scheduling memiliki nilai minimal Latency yang lebih baik sebesar 19,25 menit. Sedangkan nilai Throughput pada Fair Share Scheduling lebih cepat sebesar 0,47 Mbit/s.
English Abstract
Hadoop is a framework based on an open-source which is used for a data processing in a large scale and stored in a group of interconnected computers that are connected in a network which are distributed. Hadoop there are several scheduling algorithms to set the incoming job queue, jobs that have been submitted by the user will compete for a resource available. So the need for a scheduling algorithm in charge of managing the job on a resource and managing the data processing so that output issued in accordance with what is expected. This research was conducted to find out the comparison of performance of Fair Share Scheduling and Capacity Scheduling algorithm on job delivery on Hadoop multinode to know the value of job fail rate, latency, and throughput parameters. Capacity Scheduling supports the queue in a hierarchy that is able to divide the available resources in the cluster into multiple queues. Fair Share Scheduling is a fair allocation of resource allocation methods for queues for all jobs that enter the queue. Based on the results of tests that have been done, has a minimum value of Job Fail rate is better than Fair Share Scheduling of 0.99%. Fair Share Scheduling has a minimum value of better Latency of 19.25 minutes. While the value of Throughput on Fair Share Scheduling faster by 0.47 Mbit/s.
| Item Type: | Thesis (Sarjana) |
|---|---|
| Identification Number: | SKR/FTIK/2018/220/051801313 |
| Uncontrolled Keywords: | Hadoop, Hadoop Multi-node, Fair Share Scheduling, Capacity Scheduling, Job Fail rate, Latency, dan Throughput |
| Subjects: | 000 Computer science, information and general works > 004 Computer science > 004.015 1 Finite mathematic |
| Divisions: | Fakultas Ilmu Komputer > Teknik Informatika |
| Depositing User: | Yusuf Dwi N. |
| Date Deposited: | 25 May 2018 02:20 |
| Last Modified: | 27 Oct 2021 04:07 |
| URI: | http://repository.ub.ac.id/id/eprint/11060 |
Preview |
Text
Bagian Depan.pdf Download (1MB) | Preview |
Preview |
Text
Daftar Pustaka.pdf Download (483kB) | Preview |
Preview |
Text
BAB II.pdf Download (1MB) | Preview |
Preview |
Text
BAB I.pdf Download (508kB) | Preview |
Preview |
Text
BAB III.pdf Download (678kB) | Preview |
Preview |
Text
BAB IV.pdf Download (2MB) | Preview |
Preview |
Text
BAB V.pdf Download (1MB) | Preview |
Preview |
Text
BAB VI.pdf Download (463kB) | Preview |
Preview |
Text
Lampiran.pdf Download (748kB) | Preview |
Actions (login required)
 |
View Item |