西安电子科技大学学报 ›› 2021, Vol. 48 ›› Issue (6): 40-47.doi: 10.19665/j.issn1001-2400.2021.06.006

• 智能嵌入式系统结构与软件关键技术专栏 • 上一篇    下一篇

面向雾计算的嵌入式异构计算服务放置策略

刘锦辉(),易必杰(),张昊()   

  1. 西安电子科技大学 计算机科学与技术学院,陕西 西安 710071
  • 收稿日期:2021-08-20 出版日期:2021-12-20 发布日期:2022-02-24
  • 作者简介:刘锦辉(1984—),男,副教授,博士,E-mail: jhliu@mail.xidian.edu.cn|易必杰(1998—),男,西安电子科技大学硕士研究生,E-mail: 401599967@qq.com|张 昊(1997—),男,西安电子科技大学硕士研究生,E-mail: 844821239@qq.com
  • 基金资助:
    国家自然科学基金(61972302);国家自然科学基金(61962019);陕西省重点研发计划(2021ZDLGY07-01);陕西省重点研发计划(2021GY-014)

Embedded heterogeneous computing service placement strategy for fog computing

LIU Jinhui(),YI Bijie(),ZHANG Hao()   

  1. School of Computer Science and Technology,Xidian University,Xi’an 710071,China
  • Received:2021-08-20 Online:2021-12-20 Published:2022-02-24

摘要:

受限于云与端设备间的长距离通信,仅在云上处理数据已经不能满足时间敏感应用的需求,这促使部分应用开始向下层边缘设备扩展。随着嵌入式系统的快速发展,雾计算成为了连接云与端设备的新型计算范式,可以在更靠近数据源的地方执行应用程序。雾层计算能力通常源于高性能异构嵌入式板卡,服务在雾计算层不同的映射放置策略,对雾计算层设备的资源利用率影响巨大。现有的服务放置策略多数以提高系统服务质量为目标,忽略了嵌入式设备的异构性及计算资源的受限性,从而导致资源利用率降低。针对上述问题,提出了一种面向雾计算应用的服务放置策略,基于微服务架构,对雾计算层异构资源进行优化建模,细化了资源异构属性表征,在保证系统的基础上,通过动态对比服务放置消耗提升系统资源利用率。将这种策略与基于请求率的放置策略和iFogSim默认放置策略进行对比,该策略的系统资源利用率分别提高了约10.7%和28.7%。

关键词: 嵌入式架构, 异构资源, 容器, 雾计算, 微服务放置

Abstract:

Limited by the long distance communication between the cloud and the end device,processing data only on the cloud can no longer meet the needs of time-sensitive applications,which prompts some applications to expand to the lower edge devices.With the rapid development of embedded systems,fog computing has become a new computing paradigm that connects cloud and end devices to execute applications closer to data sources.The computing capability of the fog layer is usually derived from high-performance heterogeneous embedded board.Different mapping and placement strategies of services in the fog layer have a great impact on resource utilization of devices in the fog layer.Most of the existing service placement strategies aim at improving the system Quality of Service (QoS),but ignore the heterogeneity of embedded devices and the limitation of computing resources,which leads to the decrease in resource utilization.To solve the above problems,this paper proposes a service placement strategy for fog computing applications.Based on the micro-service architecture,the heterogeneous resources in the fog computing layer are optimized and modeled,and the heterogeneous resource attribute characterization is refined.On the basis of ensuring the system,the system resource utilization rate is improved through dynamic comparison of service placement consumption.Comparing the proposed strategy with both the request rate-based placement strategy and the iFogSim default placement strategy,the system resource utilization of the proposed strategy increases by 10.7% and 28.7%,respectively.

Key words: embedded architecture, heterogeneous resources, containers, fog computing, microservice placement

中图分类号: 

  • TP301