Scholarlink

Cloud computing (CC) has become a key paradigm in the field of information technology, providing scalable and on demand resource allocation to satisfy the dynamic needs of modern applications. Accurate workload prediction is crucial for addressing dynamic resource allocation and scheduling challenges within cloud computing environment. Although machine learning (ML) and deep learning (DL) approaches are frequently used for workload prediction, their centralized nature increases the risk of data leakage. So, federated learning overcomes the above issues. Federated learning is privacy preserving machine learning paradigm which follows the concept of moving the code to the data rather than moving the data to the code. In this paper, we use Alibaba cluster trace v2018, Alibaba cluster trace 2020, google cloud trace (GCT) 2011, and GCT 2019 for predicting the workload in cloud computing environment by applying federated leaning. The results after applying federated learning to train various deep learning models show high accuracy meanwhile preserving the security and privacy of the data.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Toward Federated Learning Approach for Workload Prediction in Cloud Computing

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Ruchika Malhotra,
  • Anjali Bansal,
  • Marouane Kessentini

摘要

Cloud computing (CC) has become a key paradigm in the field of information technology, providing scalable and on demand resource allocation to satisfy the dynamic needs of modern applications. Accurate workload prediction is crucial for addressing dynamic resource allocation and scheduling challenges within cloud computing environment. Although machine learning (ML) and deep learning (DL) approaches are frequently used for workload prediction, their centralized nature increases the risk of data leakage. So, federated learning overcomes the above issues. Federated learning is privacy preserving machine learning paradigm which follows the concept of moving the code to the data rather than moving the data to the code. In this paper, we use Alibaba cluster trace v2018, Alibaba cluster trace 2020, google cloud trace (GCT) 2011, and GCT 2019 for predicting the workload in cloud computing environment by applying federated leaning. The results after applying federated learning to train various deep learning models show high accuracy meanwhile preserving the security and privacy of the data.