Docker Enabled Virtualized Nanoservices for Local IoT Edge Networks
Docker Enabled Virtualized Nanoservices for Local IoT Edge Networks
复制标题
Docker 为本地物联网边缘网络启用虚拟化纳米服务
DOI：
--
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
M. Ylianttila
中科院分区：
文献类型：
--
作者：
J. Islam;E. Harjula;T. Kumar;Pekka Karhula;M. Ylianttila
边缘计算是一种新型计算范式，它将服务器资源更靠近终端设备。它有助于在超低延迟和传输速率方面释放高性能接入网络的全部潜力，并提高对核心网络和数据中心问题的恢复能力。多接入边缘计算（MEC）是欧洲电信标准协会（ETSI）针对接入网络级边缘计算的一种标准解决方案。MEC在接入网络层面运行，在大多数情况下是一种理想的解决方案。然而，仍然存在一些需要解决的挑战：首先与易受接入网络问题影响有关，其次是关于对接入网络和MEC服务器造成的高负载。这在大规模物联网（IoT）用例中是一个特别的问题，在这些用例中，众多传感器可能会产生大量数据，或者在接入网络出现问题时也必须确保关键系统功能。在本文中，我们研究了将一些边缘功能作为利用本地硬件能力的虚拟化且可动态部署的组件带到本地层面的可行性。为了进行这项研究，我们基于本地微服务（称为纳米服务）实现了一个本地边缘网络原型，这些纳米服务使用Docker容器实现，并使用基于Docker Swarm的编排进行部署。由于物联网网络通常由容量受限的设备组成，我们的重点是优化所提出的纳米服务的资源。
Edge computing is a novel computing paradigm moving server resources closer to end-devices. It helps unleashing the full potential of high-performance access networks with respect to ultra-low latency and transfer rate and improve resilience to problems at core networks and data centers. Multi-access Edge Computing (MEC), is a standard solution by European Telecommunications Standards Institute (ETSI) for access network-level edge computing. MEC, operating at access network level, is an ideal solution for the most cases. However, there are still some challenges to address: first is related to the vulnerability to access network problems and the second is about the high load inflicted to access networks and MEC servers. This is a particular issue in massive-scale Internet of Things (IoT) use cases, where numerous sensors may produce high amounts of data, or where critical system functionalities must be ensured also during access network problems. In this paper, we study the feasibility of bringing some edge functions to the local level as virtualized and dynamically deployable components utilizing local hardware capacity. For the study, we have implemented a local edge networking prototype based on local microservices, called nanoservices, implemented using Docker containers and deployed using Docker Swarm-based orchestration. Since IoT networks typically consist of constrained-capacity devices, our focus is in optimizing the resources of the proposed nanoservices.