Analisa Perbandingan Original Hadoop Cluster Dan Modifikasi Hadoop Cluster

Iqbal Grady Favian

Abstract


Abstract. Hadoop is an open-source framework. Various improvements may be applied to improve the performance of hadoop for processing data and stabilizing clusters. Stability is definitely required when distributions of blocks in clusters are not in order to keep the clusters stable. The experiments were in two system application versions, i.e. :original and modified. The sizes of files studied in the process were 1, 2, 3, and 4 GB resulting increase of write time speeds of the original version, as basis of comparison, compared to the ones of the modified version by 8.261 seconds, 25.7294 seconds, 9.49695, and 8.8813 seconds. On the other hands, the increases of read time speeds were by 0.1229 seconds, 2.0566 seconds, 24.3564 seconds, and 1.7612 seconds. The difference of average speed time between the original balancer and modified balancer was 3.7 minutes. Based on the results of the data analysis, it showed that block size affected read time speed and write time speed as well as balancer speed of hadoop cluster

Keywords: Hadoop, Balancer, block size.

Abstrak. Hadoop merupakan sebuah framework yang bersifat open-source maka berbagai cara dapat diterapkan untuk meningkatkan performa hadoop untuk pengolahan data dan stabilitas cluster. Stabilitas diperlukan ketika distribusi block pada cluster tidak merata sehingga cluster tetap dalam keadaan stabil. Berdasarkan hasil uji coba yang telah dilakukan dalam 2 penerapan sistem yaitu secara original dan secara modifikasi. Pada prosesnya digunakan file dengan ukuran 1, 2, 3, dan 4 GB dengan hasil pertambahan kecepatan write time antara original sebagai dasar dan modifikasi sebesar 8.261 detik, 25.7294 detik, 9.49695, dan 8.8813 detik. Dan pada kecepatan read time sebesar 0.1229 detik, 2.0566 detik, 24.3564 detik, dan 1.7612 detik. Sedangkan perbedaan waktu rata-rata kecepatan balancer original dan balancer modifikasi sebesar 3.7 menit. Berdasar pada hasil data bahwa block size dapat mempengaruhi kecepatan read time, write time dan balancer pada hadoop cluster.

Kata Kunci : Hadoop, Balancer, block size.


Full Text:

PDF

References


Chuck Lam. (2011). Hadoop In Action. Stamford: Mainning Publications Co.

Colin White. (2012). MapReduce and the Data Scientist. BI Research

Dima May. (2012). Hadoop Distributed File System (HDFS) Overview. coreservlets.com.

Holmes, A. (2012). Hadoop in Practice. New York: Manning Publications Co

White, T. (2012). Hadoop: The Definitive Guide (3rd ed.). O’Reilly Media, Inc

Priagung, Khusumanegara. (2014). Analisis Performa Kecepatan Mapreduce Pada Hadoop Menggunakan Tcp Packet Flow Analysis. Skripsi. Depok : Universitas Indonesia

Nchimbi Edward Pius, Liu Qin, Fion Yang, Zhu Hong Ming.(2012). Optimizing Hadoop Block Placement Policy & Cluster Blocks Distribution. International Journal of Computer, Electrical, Automation, Control and Information Enginering.


Refbacks

  • There are currently no refbacks.