RADL: a resource and deadline-aware dynamic load-balancer for cloud tasks

Cloud service providers acquire the computing resources and allocate them to their clients. To effectively utilize the resources and achieve higher user satisfaction, efficient task scheduling algorithms play a very pivotal role. A number of task scheduling technique have been proposed in the litera...

Full description

Saved in:
Bibliographic Details
Published in:The Journal of supercomputing 2022-08, Vol.78 (12), p.14231-14265
Main Authors: Nabi, Said, Aleem, Muhammad, Ahmed, Masroor, Islam, Muhammad Arshad, Iqbal, Muhammad Azhar
Format: Article
Language:English
Subjects:
Algorithms
Cloud computing
Compilers
Computer Science
Deadlines
Dynamic loads
Interpreters
Load balancing
Processor Architectures
Programming Languages
Resource utilization
Response time
Run time (computers)
Scheduling
Task scheduling
User satisfaction
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Cloud service providers acquire the computing resources and allocate them to their clients. To effectively utilize the resources and achieve higher user satisfaction, efficient task scheduling algorithms play a very pivotal role. A number of task scheduling technique have been proposed in the literature. However, majority of these scheduling algorithms fail to achieve efficient resource utilization that causes them to miss tasks deadlines. This is because these algorithms are not resource and deadline-aware. In this research, a Resource and deadline Aware Dynamic Load-balancer (RADL) for Cloud, tasks have been presented. The proposed scheduling scheme evenly distribute the incoming workload of compute-intensive and independent tasks at run-time. In addition, RADL approach has the capability to accommodate the newly arrived tasks (with shorter deadlines) efficiently and reduce task rejection. The proposed scheduler monitors/updates the task and VM status at run-time. Experimental results show that the proposed technique has attained up to 67.74%, 303.57%, 259.2%, 146.13%, 405.06%, and 259.14% improvement for average resource utilization, meeting tasks deadlines, lower makespan, task response time, penalty cost, and task execution cost respectively as compared to the state-of-the-art tasks scheduling heuristics using three benchmark datasets.
ISSN:0920-8542
1573-0484
DOI:10.1007/s11227-022-04426-2