What is it about?
GROUP is an end-to-end multi-step-ahead workload prediction approach focusing on workload group behavior. Because the container becomes the basic runtime unit of cloud applications, which makes workload changes no longer just an individual behavior of each container, but a group behavior of multiple containers. However, existing methods mainly focus on individual behavior. A few methods also consider the group behavior, but they only consider the original or similar data and employ the implicit regression methods. Therefore, we define the workload group behavior and its evolution and establish a workload prediction model that focuses on the workload group behavior.
Featured Image
Photo by YongGuang Tian on Unsplash
Why is it important?
The proposed workload prediction method GROUP implements the shift of workload prediction focus from individual to group, the shift of workload group behavior representation from data similarity to data correlation, and the shift of workload group behavior evolution from implicit modeling to explicit modeling. It provides a new solution for workload prediction of cloud applications, and the advantages of GROUP are proved by experiments on public datasets.
Perspectives
I think it is a great pleasure to write this paper. Because based on my basic ideas, I constantly keep thinking and discussing with my mentor. Finally, we have a complete solution and paper that is presented to a wide range of readers.
Binbin Feng
Tongji University
Read the Original
This page is a summary of: GROUP: An End-to-end Multi-step-ahead Workload Prediction Approach Focusing on Workload Group Behavior, April 2023, ACM (Association for Computing Machinery),
DOI: 10.1145/3543507.3583460.
You can read the full text:
Contributors
The following have contributed to this page
