Dynamic Resource Management of SDN-based Network Systems for IoT

随着物联网(IoT)技术将空前数量的设备连接到网络边缘，手动调配按需带宽和资源以满足这些技术的服务质量(QoS)是不可行的任务。相反，使用虚拟环境中的计算资源(也称为网络功能虚拟化(NFV))可以在网络边缘更好地管理它们。IT虚拟化和基于云的服务提供了前所未有的灵活性和敏捷性，正在改变整个行业。客户现在不再拥有自己的设备，而是可以从谷歌、亚马逊、微软和其他公司等不同的提供商那里租用或租赁计算服务。软件定义网络(SDN)是为这些托管虚拟环境提供网络服务的可编程范例。目前，许多网络功能(例如防火墙、安全设备和负载均衡器)使用位于虚拟环境之外的硬件设备；然而，使用NFV来部署这些硬件功能是很自然的，这被称为网络功能虚拟化(NFV)。这是一个关键的新兴领域，在根据应用和/或流量分析按需部署或迁移新的网络服务和功能方面具有前所未有的灵活性和敏捷性。该提案的主要目标是研究和开发方法、算法和技术，为基于物联网技术虚拟化的下一代互联网服务提供可扩展、高能效的网络基础设施。我们的重点是在以下主题上创新和开发方法：1)研究如何跨多个软件元素或虚拟机(VM)动态网络功能分解和执行传统上在专用硬件设备中实施的网络功能的方法。研究为物联网NFV部署分配网络和云资源的方法；制定机制以确保此类基于NFV的服务架构的可靠服务和服务级别保证；2)开发方法来表征工作负载、检测流量和服务异常以及3)数据中心网络(DCN)架构：研究连接和保证DCNS中的网络链路和带宽资源的方法和架构，以高效部署基于NFV的服务；加拿大始终处于电信和数据网络技术的前沿。这项研究将有助于使用最新的虚拟化技术高效部署网络服务的未来网络架构。该研究还将在未来网络架构和服务的关键领域培训HQP。

As Internet-of-Things (IoT) technology is attaching an unprecedented number of devices to the edge of the network, manual provisioning of on-demand bandwidth and resources to meet Quality-of-Service (QoS) for these technologies is not a feasible task. Instead, they are better managed at the edge of the network using compute resources within the virtual environment, also known as Network Function Virtualization (NFV). IT virtualization along with cloud-based services is transforming the industry by providing unprecedented flexibility and agility. Instead of owning their own equipment, customers can now rent or lease compute services from various providers such as Google, Amazon, Microsoft and others. Software Defined Networking (SDN) is a programmable paradigm that provides network services to these hosted virtual environments. Currently many network functions (e.g., firewalls, security appliances and load balancers) use hardware devices that are located outside of the virtual environment; however, it would be natural to use NFV to deploy these hardware functions, which is known as Network Function Virtualization (NFV). This is a key emerging area that allows unprecedented flexibility and agility in deploying or migrating new network services and functions on-demand and based on application and/or traffic profiling. The main goals of this proposal are to investigate and develop methods, algorithms and techniques to provide scalable, energy efficient network infrastructure for the next generation Internet services based on virtualization for IoT technologies. Our specific focus is to innovate and develop methods on the following topics: 1) Investigate methods on how to decompose and execute network functions that are traditionally implemented in dedicated hardware appliances across several software elements or Virtual Machines (VMs) dynamic network functions. Investigate methods to allocate network and cloud resources for IoT NFV deployment; Develop mechanisms to ensure reliable services and service level guarantees for such NFV based service architectures; 2) Develop methods to characterize workloads, detect traffic and service anomalies and 3) Data center network (DCN) architectures: Investigate methods and architectures to connect and guarantee network link and bandwidth resources in DCNs to efficiently deploy NFV based services; Canada has always been on the forefront of telecommunications and data networks technology. This research will contribute to future network architectures with efficient deployment of network services using latest virtualization technologies. The research will also train HQP in the critical area of future network architectures and services.