Please use this identifier to cite or link to this item: https://hdl.handle.net/10356/184181
Title: Evaluating job scheduling algorithms in clouds
Authors: Dixit, Ayushman
Keywords: Computer and Information Science
Issue Date: 2025
Publisher: Nanyang Technological University
Source: Dixit, A. (2025). Evaluating job scheduling algorithms in clouds. Final Year Project (FYP), Nanyang Technological University, Singapore. https://hdl.handle.net/10356/184181
Project: CCDS24-0122
Abstract: This research investigates and compares multiple job scheduling algorithms in cloud computing environments, with a strong emphasis on cost efficiency and resource optimization. Using authentic workload traces from the Google Cluster Data, the study first benchmarks the Stratus algorithm, an established cost-aware method, against baseline heuristics like Best Fit and First Fit. Experimental results show that Stratus reduces operational costs by up to 22.3% compared to these simpler approaches, while sustaining high CPU and memory utilization. Building on Stratus’s success, the research develops and evaluates several reinforcement learning (RL) algorithms, integrating advanced features such as multi-objective networks, prioritized experience replay, and proactive task migration. The best-performing RL model achieves an additional 4.2% cost reduction beyond Stratus and exhibits superior resource consolidation. Finally, a novel hybrid algorithm is proposed, merging Stratus’s heuristic advantages like runtime-aware binning and dynamic instance scaling with RL-based insights such as attention-driven feature processing and adaptive migrations. This hybrid approach further reduces costs by approximately 9.1% compared to Stratus, while maintaining high CPU and memory utilization. The findings underscore the potential of combining heuristics with data-driven learning to achieve robust, cost-effective, and resource-efficient job scheduling in modern cloud infrastructures.
URI: https://hdl.handle.net/10356/184181
Schools: College of Computing and Data Science 
Fulltext Permission: restricted
Fulltext Availability: With Fulltext
Appears in Collections:CCDS Student Reports (FYP/IA/PA/PI)

Files in This Item:
File Description SizeFormat 
Ayushman_FYP_Final_Report.pdf
  Restricted Access
3.13 MBAdobe PDFView/Open

Page view(s)

30
Updated on May 7, 2025

Download(s)

1
Updated on May 7, 2025

Google ScholarTM

Check

Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.