Scholarlink

<p>While edge computing offers lower latency, cloud computing guarantees excellent availability and cost reductions. An increasing number of enterprise application providers are now deploying applications within a cloud-edge collaborative infrastructure by using containerized microservices. Users primarily focus on the response time and availability of applications, whereas application providers strive to minimize deployment costs as much as possible. The deployment of applications in a collaborative infrastructure that combines both hybrid cloud and edge computing is defined as a multi-objective optimization problem that falls under the category of NP-hard. This paper presents DP-GA, a genetic algorithm derived from an improved version of NSGA-II, as a solution to this problem. Our strategy is to achieve a balance between lowering deployment costs and average response time, while also adhering to availability limitations. Utilizing a real dataset from Shanghai Telecom and employing the K-means clustering algorithm, we identified multiple edge data centers in Shanghai. Experimental results indicate that the proposed DP-GA method outperforms existing approaches, reducing the average response time by approximately 35% and lowering deployment costs by 15%.</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

Availability-constrained microservice deployment in hybrid cloud-edge infrastructure

  • 期刊论文

  • 发表日期：

    2025年9月3日

查看全文
  • Wei Xu,
  • Bing Tang,
  • Yujun Cao,
  • Qing Yang

摘要

While edge computing offers lower latency, cloud computing guarantees excellent availability and cost reductions. An increasing number of enterprise application providers are now deploying applications within a cloud-edge collaborative infrastructure by using containerized microservices. Users primarily focus on the response time and availability of applications, whereas application providers strive to minimize deployment costs as much as possible. The deployment of applications in a collaborative infrastructure that combines both hybrid cloud and edge computing is defined as a multi-objective optimization problem that falls under the category of NP-hard. This paper presents DP-GA, a genetic algorithm derived from an improved version of NSGA-II, as a solution to this problem. Our strategy is to achieve a balance between lowering deployment costs and average response time, while also adhering to availability limitations. Utilizing a real dataset from Shanghai Telecom and employing the K-means clustering algorithm, we identified multiple edge data centers in Shanghai. Experimental results indicate that the proposed DP-GA method outperforms existing approaches, reducing the average response time by approximately 35% and lowering deployment costs by 15%.