OpenVolcano: An Open-Source Software Platform for Fog Computing R. Bruschi * , P.Lago *‡ , G. Lamanna § , C. Lombardo *‡ , S. Mangialardi * ‡ DITEN - University of Genoa - Genoa, Italy * CNIT - University of Genoa Research Unit - Genoa, Italy § INFOCOM S.R.L. - Genoa, Italy Abstract—In order to overcome the cloud service performance limits, the INPUT Project aims to go beyond the typical IaaS-based service models by moving computing and storage capabilities from the datacenters to the edge network, and consequently moving cloud services closer to the end users. This approach, which is compatible with the concept of fog computing, will exploit Network Functions Virtualization (NFV) and Software Defined Networking (SDN) to support personal cloud services in a more scalable and sustainable way and with innovative added-value capabilities. This paper presents OpenVolcano, the open-source software platform under development in the INPUT Project, which will realize the fog computing paradigm by exploiting in-network programmability capabilities for off-loading, virtualization and monitoring. Keywords-fog computing; SDN; NFV; personal cloud services. I. INTRODUCTION The present-day Internet is characterized by a number of devices and communication media that certainly were not envisaged in the original network design. In particular, the advent of the Internet of Things (IoT) will further extend the plethora of available connected objects, bringing their number, according to Ericsson, up to 50 billion in 2020 [1]. Since such devices are characterized by limited storage and computational capabilities, their diffusion has led to the development of a significant number of cloud services to provide the support needed by smart devices. As a consequence, new networking and IT architectures are required that also take into account a more energy/cost efficient and environmental friendly approach. The joint adoption of the cloud computing supported by Network Functions Virtualization (NFV) [2] and Software Defined Networking (SDN) [3] architectures is considered a promising solution to promote flexibility and foster the introduction of new services otherwise unfeasible. This solution represents the foundation to the Infrastructure-as- a-Service (IaaS), a realization of the cloud computing paradigm that provides virtualized computing resources over the Internet. However, the proper deployment of similar paradigms is limited by the performance level currently offered by datacenters, which are located near the core networks and present end-to-end latencies around two orders of magnitude higher than the required values. To address this issue, fog computing has been proposed [4] by CISCO with the goal of deploying storage, computing and configuration features closer to the end-user instead of inside datacenters. This deployment scheme has three main advantages: i) the reduced distance of the user from his/her services reduces the end-to-end latency, hence improving overall Quality of Services and Experience (QoS and QoE); ii) the deployment of services inside the edge network allows for better control on the allocation of physical and logical resources, with the possibility of applying economies of scale and power management schemes, and improved privacy; iii) additionally, the telecom providers can hide the complexity of the underlying network infrastructure, and hence promote simplified applications design. In this respect, the INPUT Project [5] is devoted to foster Future Internet infrastructures beyond the typical IaaS- based service models by moving computing and storage capabilities, from both cloud services and user devices, to the edge network. This approach will exploit the ability to directly access network primitives, and will improve scalability in the interactions of the network with users and datacenters. In order to provide scalable and virtualized networking technologies able to natively integrate cloud services, both personal and federated, the INPUT Project is realizing OpenVolcano, an open-source software platform for fog computing, which will exploit in-network programmability capabilities for off-loading, virtualization and monitoring. The paper is organized as follows. Section II focuses on the deployment of the personal cloud services. Section III reports the rationale behind the OpenVolcano architecture, while its control and data plane building blocks are described in Sections IV and V, respectively. Section VI provides further details and performance evaluation on quake, the internal virtual switch. Finally, conclusions are drawn in Section VII. II. DEPLOYMENT OF PERSONAL CLOUD SERVICES The INPUT technologies will enable next-generation cloud applications to go beyond classical service models and even to replace physical Smart Devices (SD), usually placed in users’ homes (e.g., network-attached storage servers, set-top-boxes, video recorders, home automation control units, etc.) or deployed around for monitoring purposes (e.g., sensors), with their “virtual images,” providing them to users “as a Service”. A virtual image is defined to be a software instance that dematerializes a physical network-connected device, and that provides its virtual presence in the network and all its functionalities. Virtual images are meant to realize smarter, always and everywhere accessible, performance-unlimited virtual devices into the cloud. SDs can be fully or partially virtualized depending on their constraints on physical components: sensors cannot be entirely virtualized, since they still need the hardware to acquire measures and to transfer them elsewhere. On the contrary, entertainment appliances can be fully virtualized, and interfaced with visualization devices (e.g., a television or a tablet) through standard and well-known protocols. A breakdown of the 2016 28th International Teletraffic Congress 978-0-9883045-1-2/16 $31.00 © 2016 ITC 22 2016 28th International Teletraffic Congress 978-0-9883045-1-2/16 $31.00 © 2016 ITC 22