ANALISIS KOMPARASI PERFORMA LAYANAN SERVER-BASED DAN SERVERLESS DI GOOGLE CLOUD PLATFORM (GCP) (STUDI KASUS: Deployment Model Machine Learning)

Vina Fujiyanti, - (2024) ANALISIS KOMPARASI PERFORMA LAYANAN SERVER-BASED DAN SERVERLESS DI GOOGLE CLOUD PLATFORM (GCP) (STUDI KASUS: Deployment Model Machine Learning). S1 thesis, Universitas Pendidikan Indonesia.

Baca Full Text klik disini
[img] Text
S_SISTEL_2005751_Title.pdf
Download (866kB)
[img] Text
S_SISTEL_2005751_Chapter1.pdf
Download (58kB)
[img] Text
S_SISTEL_2005751_Chapter2.pdf
Restricted to Staf Perpustakaan
Download (714kB)
[img] Text
S_SISTEL_2005751_Chapter3.pdf
Download (386kB)
[img] Text
S_SISTEL_2005751_Chapter4.pdf
Restricted to Staf Perpustakaan
Download (935kB)
[img] Text
S_SISTEL_2005751_Chapter5.pdf
Download (39kB)
[img] Text
S_SISTEL_2005751_Appendix.pdf
Restricted to Staf Perpustakaan
Download (7MB)
Official URL: https://repository.upi.edu/

Abstract

Penggunaan teknologi cloud computing menjadi pertimbangan para pengguna karena biayanya yang efisien, pengelolaan yang terpusat, fleksibel, dan layanannya yang bervariasi, salah satunya layanan komputasi. Penyedia infrastruktur cloud seperti Google Cloud Platform (GCP) menyediakan layanan komputasi server-based dan serverless untuk deployment aplikasi seperti model machine learning. Baik layanan server-based dan serverless memiliki karakteristik dan kelebihannya masing-masing yang membuat pengguna kesulitan dalam memilih layanan yang sesuai dengan kebutuhan. Oleh karena itu, penelitian ini dilakukan untuk mengkomparasi layanan server-based dan serverless dengan melakukan deployment model machine learning untuk mengetahui layanan terbaik berdasarkan pengukuran performa layanan. Performa layanan yang akan diukur pada penelitian ini adalah CPU dan memory utilization, latency, pricing, dan parameter tambahan berdasarkan developer experiences yang meliputi kompatibilitas framework machine learning, tingkat kemudahan penerapan, dan tingkat ketersediaan dokumentasi. Pengujian dilakukan dengan mengirim 100 kali permintaan terhadap endpoint hasil deployment melalui HTTP request GET pada JMeter. Hasil penelitian menunjukkan bahwa layanan Cloud Run terbukti ringan dan unggul untuk melakukan deployment berdasarkan pengukuran performa karena memiliki utilisasi resource yang rendah yaitu CPU utilization sebesar 0,05%, memory utilization sebesar 0,91%, latency sebesar 2,70 ms, serta biaya yang efisien sebesar Rp 6.493. Penelitian ini menunjukkan bahwa layanan serverless memiliki performa yang lebih baik dan optimal untuk melakukan deployment, khususnya pada model machine learning. ----- The use of cloud computing technology is considered by users because of its cost efficiency, centralized management, flexibility, and variety of services, including computing services. Cloud infrastructure providers such as Google Cloud Platform (GCP) provide server-based and serverless computing services for application deployment such as machine learning models. Both server-based and serverless services have their own characteristics and advantages that make it difficult for users to choose the service that suits their needs. Therefore, this research is conducted to compare server-based and serverless services by deploying machine learning models to determine the best service based on service performance measurements. The service performance that will be measured in this research is CPU and memory utilization, latency, pricing, and additional parameters based on developer experiences which include machine learning framework compatibility, ease of implementation, and level of documentation availability. Testing is done by sending 100 requests to the endpoint through HTTP request GET on JMeter. The results showed that the Cloud Run service proved to be lightweight and superior for deployment based on performance measurements because it has low resource utilization, namely CPU utilization of 0.05%, memory utilization of 0.91%, latency of 2.70ms, and an efficient cost of Rp 6,493. This research shows that serverless services have better and optimal performance for deployment, especially in machine learning models.

Item Type: Thesis (S1)
Additional Information: https://scholar.google.com/citations?view_op=new_profile&hl=id ID SINTA Dosen Pembimbing: Galura Muhammad Suranegara: 6703764 Ichwan Nul Ichsan: 6721201
Uncontrolled Keywords: GCP, server-based, serverless, deployment, machine learning GCP, server-based, serverless, deployment, machine learning
Subjects: T Technology > T Technology (General)
Divisions: UPI Kampus Purwakarta > S1 Sistem Telekomunikasi
Depositing User: Vina Fujiyanti
Date Deposited: 08 Aug 2024 07:35
Last Modified: 08 Aug 2024 07:35
URI: http://repository.upi.edu/id/eprint/119999

Actions (login required)

View Item View Item