Large scale analytics for workload segmentation

Loading...
Thumbnail Image
Authors
Hu, Bing
Mason, Nicholas
Advisors
Issue Date
2023-06
Type
Article
Keywords
Workload characterization , Clustering , Interpretable classifier , Time-phased analysis , Machine learning , Statistics , Benchmark workload , Big data
Research Projects
Organizational Units
Journal Issue
Citation
Hu, B. & Mason, N. (2023). Large scale analytics for workload segmentation. Journal of Management & Engineering Integration, 16(1), 19-26. https://doi.org/10.62704/10057/25984
Abstract

As the complexity and size of workload landscapes continue to evolve, traditional tools that are useful for characterizing workloads have gradually become inadequate. Approaches such as scaling studies or isolated analyses of a single workload are gradually becoming insufficient to understand the broad state of the workload universe. To address this problem, we follow a machine learning methodology to leverage large numbers of workload experiments that already exist on Intel. We advanced a workload analysis platform that stores, manages, and facilitates the analysis of workload information using two big data analysis software tools, a top-down workload decision classifier and a workload universe mapping tool, to characterize large-scale workloads. Both analysis tools present a novel way to look at workload data from a higher level and consider these new characteristics at the workload segmentation level. We seek to achieve this goal in alignment with the strategic goals of developing new benchmarks, informing market sizing, and detecting emerging workloads.

Table of Contents
Description
Published in SOAR: Shocker Open Access Repository by Wichita State University Libraries Technical Services, October 2023.
Publisher
Association for Industry, Engineering and Management Systems (AIEMS)
Journal
Book Title
Series
Journal of Management & Engineering Integration
v.16 no.1
PubMed ID
ISSN
1939-7984
EISSN