Implementasi Load Balancing Web Server dengan Metode Weighted Response Time pada Software Defined Network (SDN) Pembimbing: Ir. Heru Nurwasito, M.Kom

Rahmawati, Risna (2021) Implementasi Load Balancing Web Server dengan Metode Weighted Response Time pada Software Defined Network (SDN) Pembimbing: Ir. Heru Nurwasito, M.Kom. Sarjana thesis, Universitas Brawijaya.

Abstract

Banyaknya akses internet yang terjadi dalam beberapa tahun ini menyebabkan beberapa layanan jaringan membutuhkan sebuah skema load balancing. Load balancing pada jaringan tradisional membutuhkan load balancer dengan dedicated hardware yang membutuhkan biaya besar dan mudah terjadi single point failure. Pada jaringan Software Defined Network, load balancer bersifat programmable. Penelitian sebelumnya membahas mengenai penerapan salah satu algoritma load balancing pada Software Defined Network berupa algoritma response time. Algoritma load balancing response time membagi beban berdasarkan response time server, akan tetapi algoritma tersebut tidak memperhitungkan spesifikasi yang dimiliki server. Sedangkan, dalam penerapan layanan internet memungkinkan jika server yang digunakan memiliki spesifikasi yang berbeda. Oleh karena itu, pada penelitian ini, diterapkan sebuah algoritma load balancing weighted response time. Algoritma weighted response time membagi beban server berdasarkan rasio nilai response time dan weight yang merepresentasikan spesifikasi server. Dengan algoritma weighted response time, permintaan client diarahkan ke server dengan rasio nilai weight dan response time yang paling rendah. Pada penelitian ini, kinerja jaringan diuji dengan parameter response time, throughput, dan CPU utilization dengan tiga skenario pengujian. Hasil pengujian menunjukkan bahwa jaringan dengan algoritma weighted response time membutuhkan waktu respon rata-rata lebih rendah dan menghasilkan nilai throughput yang lebih tinggi dibandingkan algoritma response time. Berdasarkan pengujian, jaringan dengan algoritma weighted response time memiliki nilai response time terendah yaitu sebesar 2,17 ms dan nilai throughput tertinggi yaitu sebesar 100,70 KB/s dari keseluruhan hasil pengujian. Sedangkan untuk parameter CPU utilization, jaringan dengan algoritma response time memiliki nilai yang terendah dari keseluruhan hasil pengujian yaitu sebesar 69,26%.

English Abstract

The large number of internet access that has occurred in recent years has resulted several network services are requiring a load balancing scheme. Load balancing on traditional networks requires load balancers with dedicated hardware which is costly and can easily occur single point failure. On a Software Defined Network, the load balancer is programmable. Previous research discusses the application of load balancing method based on server response time as one of the load balancing algorithms on a Software Defined Network. Load balancing method based on server response time divides server load based on the server response time, however this algorithm does not take server specifications into account for load balance. Meanwhile, in the application of internet services it is possible if the servers used have different specifications. Therefore, in this study, a load balancing weighted response time algorithm is applied. The weighted response time algorithm divides the server load based on the ratio of response time and weight values that represent server specifications. This weighted response time algorithm directed client requests to the server with the lowest ratio weight value and response time. In this study, the network performance of the weighted response time algorithm was tested with parameters of response time, throughput, and CPU utilization with three test scenarios. The test results show that the network with weighted response time algorithm requires a lower average response time and produces a higher throughput value than the network with response time algorithm. Based on the test, the network with weighted response time algorithm has the lowest response time value of 2.17 ms and the highest throughput value of 100.70 KB/s from the overall test results. As for the CPU utilization parameter, the network with response time based algorithm has the lowest value of the overall test results, which is 69.26%.

Item Type: Thesis (Sarjana)
Identification Number: 0521150124
Uncontrolled Keywords: Load Balancing, Software Define Network, Weighted Response Time, Weight Load Balancing, Software Define Network, Weighted Response Time, Weight
Subjects: 000 Computer science, information and general works > 004 Computer science
Divisions: Fakultas Ilmu Komputer > Teknik Informatika
Depositing User: Budi Wahyono Wahyono
Date Deposited: 04 Nov 2021 07:04
Last Modified: 17 Oct 2024 02:04
URI: http://repository.ub.ac.id/id/eprint/186500
[thumbnail of Risna Rahmawati..pdf] Text
Risna Rahmawati..pdf

Download (2MB)

Actions (login required)

View Item View Item