Fernando N. N. Farias

Control of Multiple Packet Schedulers for Improving QoS on OpenFlow/SDN Networking

Packet scheduling is essential to properly support applications on Software-Defined Networking (SDN) model. However, on OpenFlow/SDN, QoS is only performed with bandwidth guarantees and by a well-known FIFO scheduling. Facing this limitation, this paper presents the QoSFlow proposal, which controls multiple packet schedulers of Linux kernel and improve the flexibility of QoS control. The paper assesses QoSFlow performance, by analysing response time of packet scheduler operations running on datapath level, maximum bandwidth capacity, hardware resource utilization rate, bandwidth isolation and QoE. Our outcomes show an increase more than 48% on PSNR value of QoE by using SFQ scheduling.

A proposal management of the legacy network environment using OpenFlow control plane

The Future Internet will arise from the convergence of new network concepts and combine technologies, services, media and content. It will offer flexibility and diversity with scalable content and services that are accessible through a wide range of interfaces and devices. However, the biggest challenge now is how to enable and test the proposed approaches so that they can be validated without sacrificing the current production infrastructure. The OpenFlow protocol allows production networking environments such as campus networks, metropolitan networks or R&D networks, to be used as experimental infrastructure hosting, future Internet architectures, software and protocols, in parallel with the production traffic. During rollout, there is a practical problem that arises with Legacy networks that do not support OpenFlow and need to be replaced/upgraded or refined by means of costly network re-engineering. This paper proposes a new OpenFlow architecture with new components, capable of managing Legacy-non-OpenFlow elements by offering a new solution that facilitates the management of Legacy technologies and allows them to be employed in FI experimentation environment and increase the number of experiment with the Legacy Network Environment using OpenFlow control.

Additions to the ETArch control plane to support multimedia QoS-guaranteed content transport over OpenFlow-enabled SDN future internet systems

The Future Internet approach requires new solutions to support novel usage scenarios driven by the technological evolution and the new service demands. However, this paradigm shift requires deeper changes in the existing systems, which makes Internet providers reluctant in deploying the full transformation required for the Future Internet. The Entity Title Architecture (ETArch) is a holistic clean-slate Future Internet system embedding new services for these scenarios leveraging the Software Defined Networking (SDN) concept materialized by the OpenFlow. However, legacy ETArch deploys a fully per-flow approach to provision the same transport model for all sessions (equivalent to the Internet best-effort), while suffering with performance drawbacks and lacking Quality of Service (QoS) control. To that, we evolved ETArch with SMART (Support of Mobile Sessions with High Transport Network Resource Demand) QoS control approach, which coordinates admission control and dynamic control of super-dimensioned resources to accommodate multimedia sessions with QoS-guaranteed over time, while keeping scalability/performance and users with full Quality of Experience (QoE). The SMART-enabled ETArch system evaluation was carried out using a real Testbed of the OFELIA Brazilian Island, confirming its benefits in both data and control planes over the legacy ETArch.

On the Benchmarking Mainstream Open Software-Defined Networking Controllers

Software-Defined Networking (SDN) has been one of the most successfull networking model over the past few years. The model decouples the network control and forwarding functions enabling the underlying infrastructure complexity to be programmed by applications. Although control plane is the cern for all the benefits, it is also the most crucial drawback of the SDN model to keep up working. Therefore, this paper presents a performance analysis on mainstream open-source SDN controllers. The results show that a well-perfomed control plane not only depends on controller throughput and response time, but also relies on topology discovery time. Our results show that Beacon controller has the highest performance on controller troughput because it uses multicore feature better than others. However, there is a few difference on topology building delay when compared to ONOS, Floodlight and OpenDaylight, which are also Java-based controllers. The worst one is Ryu for building network topology.