Analisis Kinerja Delay Scheduling Pada Hadoop Terhadap Karakteristik Job Fail Rate Untuk Pengiriman Dua Job Yang Berbeda

Wahyuddin Saputra, Alvianus Dengen (Teknik Elektro, Universitas Teknologi Sulawesi)

Celebes Computer Science Journal
Vol 1 No 1 (2019): Celebes Computer Science Journal

This research will compare two algorithms in the process of scheduling large-scale data that runs on VPS (Virtual Private Server) by utilizing an open source platform, Hadoop and Proxmox as a container for creating VPS. For the research method, it will use the FIFO algorithm and the delay scheduling algorithm, each of which will run on infrastructure that has been built previously. Of the three types of scenarios that are done by using two different types of jobs in job submission, it is obtained that the delay scheduling algorithm produces a smaller fail rate than the FIFO algorithm. This can occur because of the characteristics of the delay scheduling algorithm which is to improve job locality data but the more number of jobs the job fail rate performance will increase with the appearance of failures.