DIstributed SDN COntrollers for rich and elastic network services – DISCO

SDN (Software Defined Networking) represents a disruptive change in the way networks are architected, built, and operated. SDN opens up traditionally closed, single-vendor networks and brings competition and innovation to bear on some historically unsolvable networking problems. SDN is a new promising networking paradigm where network control is decoupled from forwarding and is directly programmable. This migration of control, formerly tightly bound in individual network devices, into accessible computing devices enables the underlying infrastructure to be abstracted for applications and network services, which can treat the network as a logical or virtual entity. SDN thus enables to automatically and flexibly manage network resources.
This project proposes to explore the way how SDN changes network monitoring, control, urbanisation and abstract description of network resources for the optimisation of services. SDN promises a wide array of optimization techniques and newer algorithms for managing traffic. Over the next five years, SDN will penetrate further into Telco and data centre networks and will find its place into enterprises. However, even though the Open Network Foundation (ONF) progresses on the standardisation of an open control interface (OpenFlow), there are still many research issues requiring investigation and the whole approach calls for experimental validation.
DISCO (DIstributed SDN COntrollers for rich and elastic network services) leans on the complementary technical expertise of the partners to bring flexibility, scalability and resiliency in SDN architectures, but also, richness and elasticity for the deployment of network services, for example for content delivery. DISCO will in particular investigate the urbanisation of virtual network functions or network appliances (e.g., load-balancer, deep packet inspection, ciphers), that can be embedded into virtual machines, thus consuming communized CPUs, memory and network capabilities. The project will provide tools to optimise the management of appliances in order to have the minimal footprint on the physical layer while ensuring high level services (content distribution, critical information…). DISCO will propose innovative SDN data plane optimization mechanisms, controllable through a standard SDN interface and based on an extension of current technologies. These mechanisms will dynamically accelerate the flow processing while consolidation other workloads on a single commoditized multi-core platform. DISCO will also investigate a multi-controller platform for robustness, scalability and flexibility of the control plane and extended capabilities for monitoring and configuration of virtualised network functions. Finally, DISCO will study and develop a NaaS (Network-as-a-Service) API that provides a mean for applications to get an instantaneous abstract view of network resources to determine the amount of resources they can use and compute their optimal placement and to request network services.
DISCO will be based on a use case with two principal axes. The first axis corresponds to the management of critical applications. Critical applications require the differentiation between vital/critical flows and low-priority flows, but also the management of heterogeneous Quality of Service constraints (bandwidth, latency, jitter, etc.) on the flows. The second axis corresponds to a content distribution service. Such a service needs to have an abstract view of the underlying resources to be able to optimally place caches and program synchronisation, redirections, etc. to react rapidly to traffic changes (e.g., flash crowds).
DISCO is an appealing SDN-based project with a strong potential to reshape future network infrastructures. We expect DISCO results to impact three areas: standards and SDN communities, scientific and research. The partners have extensive experience in networks, multi-core platforms, ONF and large-scale experiments.