Seer: Empowering Software Defined Networking with Data Analytics
📝 Abstract
Network complexity is increasing, making network control and orchestration a challenging task. The proliferation of network information and tools for data analytics can provide an important insight into resource provisioning and optimisation. The network knowledge incorporated in software defined networking can facilitate the knowledge driven control, leveraging the network programmability. We present Seer: a flexible, highly configurable data analytics platform for network intelligence based on software defined networking and big data principles. Seer combines a computational engine with a distributed messaging system to provide a scalable, fault tolerant and real-time platform for knowledge extraction. Our first prototype uses Apache Spark for streaming analytics and open network operating system (ONOS) controller to program a network in real-time. The first application we developed aims to predict the mobility pattern of mobile devices inside a smart city environment.
💡 Analysis
Network complexity is increasing, making network control and orchestration a challenging task. The proliferation of network information and tools for data analytics can provide an important insight into resource provisioning and optimisation. The network knowledge incorporated in software defined networking can facilitate the knowledge driven control, leveraging the network programmability. We present Seer: a flexible, highly configurable data analytics platform for network intelligence based on software defined networking and big data principles. Seer combines a computational engine with a distributed messaging system to provide a scalable, fault tolerant and real-time platform for knowledge extraction. Our first prototype uses Apache Spark for streaming analytics and open network operating system (ONOS) controller to program a network in real-time. The first application we developed aims to predict the mobility pattern of mobile devices inside a smart city environment.
📄 Content
Seer: Empowering Software Defined Networking with Data Analytics
Kyriakos Sideris, Reza Nejabati, Dimitra Simeonidou High Performance Networks Group University of Bristol, UK kyriakos.sideris@bristol.ac.uk
Abstract–Network complexity is increasing, making network control and orchestration a challenging task. The proliferation of network information and tools for data analytics can provide an important insight into resource provisioning and optimisation. The network knowledge incorporated in software defined networking can facilitate the knowledge driven control, leveraging the network programmability. We present Seer: a flexible, highly configurable data analytics platform for network intelligence based on software defined networking and big data principles. Seer combines a computational engine with a distributed messaging system to provide a scalable, fault tolerant and real-time platform for knowledge extraction. Our first prototype uses Apache Spark for streaming analytics and open network operating system (ONOS) controller to program a network in real-time. The first application we developed aims to predict the mobility pattern of mobile devices inside a smart city environment. Keywords–Big data, data analytics, data mining, knowledge centric networking (KCN), software defined networking (SDN), Seer I. INTRODUCTION A big growth in networks has been observed in recent years, ranging from end-user mobile data networks through to machine-to-machine communications. The number of devices connected to IP networks will continue to increase up to three times the global population in 2020, while the global Internet traffic will be equivalent to 95 times the volume of the entire global Internet in 2005 [1]. As the complexity and interdependence of networks increase, management will be a progressively challenging task. The importance of data analytics has increased due to the fact that data can fuel opportunities across many disciplines. For instance, businesses can deliver personalised services in social media and e-commerce, and concurrently researchers can understand and investigate the secrets of the human genome. Data analytics inherited a plethora of statistics, machine learning and data mining algorithms, only to combine with scalable and fault tolerant computer science methodologies and finally deliver big data tools. Network orchestrators and cloud providers are enabled to optimise their services by using big data. On the other hand, the software defined networking (SDN) paradigm facilitates network evolution and innovation by simplifying network hierarchy and separating the control from data planes. The extracted network knowledge can drive intelligent SDN control by providing a clearer view of the network dynamics. The importance of network analytics increases as SDN is further established in the networking market [2]. Although various frameworks have been proposed to resolve network optimisation problems using big data analytics, they do not propose an architecture that can work as common ground for algorithm development. Moreover, reliability is becoming increasingly important as network elements become main components of the control plane. Therefore, any proposed platform should meet the challenges of high availability and scalability whilst being adaptive enough to carry out a diverse range of tasks. In addition, it should also run on commodity hardware in order to reduce the capital expenditure. In this paper we introduce Seer, a flexible, highly configurable data analytics platform for network intelligence based on SDN and big data principles. Seer uses active or passive network elements, which are either SDN enabled or not, to extract information related to the network state. Various sources of information are combined in order to generate knowledge, which in turn is used to leverage SDN, either by optimising the network resource allocation or by predicting the next state and proactively orchestrating resources. The extracted knowledge is also presented to network applications in the higher layers to assist their management processes. Our platform provides the benefits of big data architectures on network data management and data analytics [3] to the SDN network paradigm in order to leverage network programmability, ease of management, optimisation of network resources and eminent failure recovery [4]. The main improvement of Seer lies in the fact that the platform has been built on mature open-source tools, using design methodologies employed in industry. All elements of the proposed platform run on commodity hardware and can adapt to any network analytics related task. The applications running on Seer can vary from core to edge networks, targeting storage, computing or orchestration applications. The developer defines the type of extracted information from the network, the format that wil
This content is AI-processed based on ArXiv data.