A novel weight-assignment load balancing algorithm for cloud applications

Adewojo, A; Bass, J

Abstract

Web applications commonly suffer from flash crowds and resource failure, resulting in performance degradation. Flash crowds are large, sudden, yet legitimate influxes of user requests that constitute a critical problem because of their potential economic damage. For cloud providers, resource estimation is challenging, while distributing workload and sustaining performance. To alleviate flash crowds and resource failure problems, we propose a novel weight assignment load balancing algorithm that combines five carefully selected server metrics to efficiently
distribute the workload of three-tier web applications among virtual
machines. We experimentally characterised, using a private cloud running OpenStack, the load distribution ability of our proposed novel
algorithm, as well as a baseline algorithm and round-robin algorithm.
We compared the performance of the three algorithms by simulating
resource failures and flash crowds, while carefully measuring response
times. Our experimental results show that our approach improves average response times by 12.5% when compared to the baseline algorithm
and 22.3% when compared to the round-robin algorithm in the flash
crowds’ situation. In addition, average response time was improved
by 20.7% when compared to the baseline algorithm and 21.4% when
compared to the round-robin algorithm in resource failure situations.
These experiments show that our novel algorithm is more resilient
to fluctuating loads and resource failures than baseline algorithms.

Citation

Adewojo, A., & Bass, J. (in press). A novel weight-assignment load balancing algorithm for cloud applications. SN Computer Science, 4, https://doi.org/10.1007/s42979-023-01702-7

• Journal Article Type: Article
• Acceptance Date: Jan 9, 2023
• Online Publication Date: Mar 17, 2023
• Deposit Date: Jan 20, 2023
• Publicly Available Date: Mar 21, 2023
• Journal: SN Computer Science
• Publisher: Springer
• Volume: 4
• DOI: https://doi.org/10.1007/s42979-023-01702-7
• Publisher URL: https://doi.org/10.1007/s42979-023-01702-7