article · 2023

A Next Function Set Scheduler for Serverless Applications

Cited by 0Abstract and details only

Abstract

Serverless computing, a dominant cloud-based application development paradigm, allows developers to focus on logic rather than deployment concerns. Despite these benefits, the cold start problem hinders Serverless applications, impacting the performance. Existing research often addresses individual function cold starts, but real-world Serverless applications involve multiple functions and internal dependencies. Furthermore, the prediction of forthcoming dependent functions is not taken into consideration. This paper explores the cold start problem in Serverless application level, proposing Next Function Set Scheduler (NFS) to address this problem. NFS discovers dependences between Serverless functions and derive function sets based on dependences. Additionally, NFS take the prediction of forthcoming function set into consideration to achieve better cold start mitigation performance. We evaluate NFS by conduct experiments on an industrial Serverless dataset. The results show that NFS reduce 28.3% cold start rate for 75% of functions and while having a lower memory usage compared with SOTA methods.

Cite this paper

APA 7

Li, Z., Yu, X., & Chun, W. (2023). A Next Function Set Scheduler for Serverless Applications. https://doi.org/10.1109/eiecc60864.2023.10456681

MLA 9

Li, Zhuo, et al. “A Next Function Set Scheduler for Serverless Applications.” 2023, pp. 173–77, https://doi.org/10.1109/eiecc60864.2023.10456681.

Chicago

Li, Zhuo, Xinzhou Yu, and Wang Chun. 2023. “A Next Function Set Scheduler for Serverless Applications.” https://doi.org/10.1109/eiecc60864.2023.10456681.

Harvard

Li, Z., Yu, X. and Chun, W. (2023) ‘A Next Function Set Scheduler for Serverless Applications’. Available at: https://doi.org/10.1109/eiecc60864.2023.10456681.

BibTeX

@article{li2023next,
  title = {{A Next Function Set Scheduler for Serverless Applications}},
  author = {Li, Zhuo and Yu, Xinzhou and Chun, Wang},
  pages = {173--177},
  year = {2023},
  doi = {10.1109/eiecc60864.2023.10456681}
}

Made from the index record. Check names, capitalisation and page numbers against the paper. Need another style? AI for Research has over 10,000.