Scholarlink

<p>Edge computing reduces latency and alleviates cloud dependency, making it essential for modern network infrastructures. Docker facilitates containerized application deployment in edge environments, but its default limit of three concurrent image downloads per client can delay provisioning when network resources are underutilized. This paper proposes ADAPT, an adaptive approach that dynamically adjusts parallel downloads based on real-time network conditions. By increasing concurrency during low congestion and limiting it during peak usage, our method optimizes provisioning times and enhances system performance. Extensive experimental results confirm the feasibility of the proposed method and demonstrate that ADAPT consistently outperforms both the default Docker configuration and multiple state-of-the-art techniques for container image distribution. The results show provisioning time reductions of up to 33% under realistic edge computing scenarios, reinforcing ADAPT as a robust and practical solution for dynamic infrastructures.</p>

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

Improving edge application provisioning with adaptive container image management

  • 期刊论文

  • 发表日期：

    2025年9月8日

查看全文
  • Pedro Henrique Sachete Garcia,
  • Arthur Francisco Lorenzon,
  • Marcelo Caggiani Luizelli,
  • Fábio Diniz Rossi

摘要

Edge computing reduces latency and alleviates cloud dependency, making it essential for modern network infrastructures. Docker facilitates containerized application deployment in edge environments, but its default limit of three concurrent image downloads per client can delay provisioning when network resources are underutilized. This paper proposes ADAPT, an adaptive approach that dynamically adjusts parallel downloads based on real-time network conditions. By increasing concurrency during low congestion and limiting it during peak usage, our method optimizes provisioning times and enhances system performance. Extensive experimental results confirm the feasibility of the proposed method and demonstrate that ADAPT consistently outperforms both the default Docker configuration and multiple state-of-the-art techniques for container image distribution. The results show provisioning time reductions of up to 33% under realistic edge computing scenarios, reinforcing ADAPT as a robust and practical solution for dynamic infrastructures.