Traffic Engineering (TE) in IP carrier networks is one of the functions that can benefit from the Software Defined Networking paradigm. By logically centralizing the control of the network, it is possible to "program" per-flow routing based on TE goals. Traditional per-flow routing requires a direct interaction between the SDN controller and each node that is involved in the traffic paths. Depending on the granularity and on the temporal properties of the flows, this can lead to scalability issues for the amount of routing state that needs to be maintained in core network nodes and for the required configuration traffic. On the other hand, Segment Routing (SR) is an emerging approach to routing that may simplify the route enforcement delegating all the configuration and per-flow state at the border of the network. In this work we propose an architecture that integrates the SDN paradigm with SR-based TE, for which we have provided an open source reference implementation. We have designed and implemented a simple TE/SR heuristic for flow allocation and we show and discuss experimental results.

Artificial Neural Networks (NNs) play an increasingly important role in many services and applications, contributing significantly to compute infrastructures' workloads. When used in latency sensitive services, NNs are usually processed by CPUs since using an external dedicated hardware accelerator would be inefficient. However, with growing workloads size and complexity, CPUs are hitting their computation limits, requiring the introduction of new specialized hardware accelerators tailored to the task. In this paper we analyze the option to use programmable network devices, such as Network Cards and Switches, as NN accelerators in place of purpose built dedicated hardware. To this end, in this preliminary work we analyze in depth the properties of NN processing on CPUs, derive options to efficiently split such processing, and show that programmable network devices may be a suitable engine for implementing a CPU's NN co-processor. CCS CONCEPTS• Networks → Programmable networks; In-network processing; • Computing methodologies → Machine learning; • Computer systems organization → Neural networks;

Abstract-In this paper, we first introduce the NFV architecture and the use of IPv6 Segment Routing (SRv6) network programming model to support Service Function Chaining in a NFV scenario. We describe the concepts of SR-aware and SRunaware Virtual Network Functions (VNFs). The detailed design of a network domain supporting VNF chaining through the SRv6 network programming model is provided. The operations to support SR-aware and SR-unaware VNFs are described at an architectural level and in particular we propose a solution for SR-unaware VNFs hosted in a NFV node. The proposed solution has been implemented for a Linux based NFV host and the software is available as Open Source. Finally, a methodology for performance analysis of the implementation of the proposed mechanisms is illustrated and preliminary performance results are given.

Tor Vergata -(2) Consortium GARR -(3) CREATE-NET -(4) ON.Lab

The introduction of SDN in IP backbones requires the coexistence of regular IP forwarding and SDN based forwarding. The former is typically applied to best effort Internet traffic, the latter can be used for different types of advanced services (VPNs, Virtual Leased Lines, Traffic Engineering…). In this paper we first introduce the architecture and the services of an "hybrid" IP/SDN networking scenario. Then we describe the design and implementation of an Open Source Hybrid IP/SDN (OSHI) node. It combines Quagga for OSPF routing and Open vSwitch for OpenFlow based switching on Linux. The availability of tools for experimental validation and performance evaluation of SDN solutions is fundamental for the evolution of SDN. We provide a set of open source tools that allow to facilitate the design of hybrid IP/SDN experimental networks, their deployment on Mininet or on distributed SDN research testbeds and their test. Finally, using the provided tools, we evaluate key performance aspects of the proposed solutions. The OSHI development and test environment is available in a VirtualBox VM image that can be downloaded.