David Hock

Heuristic Approaches to the Controller Placement Problem in Large Scale SDN Networks

Software Defined Networking (SDN) marks a paradigm shift towards an externalized and logically centralized network control plane. A particularly important task in SDN architectures is that of controller placement, i.e., the positioning of a limited number of resources within a network to meet various requirements. These requirements range from latency constraints to failure tolerance and load balancing. In most scenarios, at least some of these objectives are competing, thus no single best placement is available and decision makers need to find a balanced trade-off. This work presents POCO, a framework for Pareto-based Optimal COntroller placement that provides operators with Pareto optimal placements with respect to different performance metrics. In its default configuration, POCO performs an exhaustive evaluation of all possible placements. While this is practically feasible for small and medium sized networks, realistic time and resource constraints call for an alternative in the context of large scale networks or dynamic networks whose properties change over time. For these scenarios, the POCO toolset is extended by a heuristic approach that is less accurate, but yields faster computation times. An evaluation of this heuristic is performed on a collection of real world network topologies from the Internet Topology Zoo. Utilizing a measure for quantifying the error introduced by the heuristic approach allows an analysis of the resulting trade-off between time and accuracy. Additionally, the proposed methods can be extended to solve similar virtual functions placement problems which appear in the context of Network Functions Virtualization (NFV).

Pareto-optimal resilient controller placement in SDN-based core networks

Recently, Software Defined Networking (SDN) has gained a lot of attention, even for the use in core communication networks. When deploying SDN in large core networks, the number and location of controllers must be carefully planned. A first study on this topic by Heller et al. [3] was followed by more detailed publications that included resilience or dynamic controller provisioning. In our previous work [2], we provide an overview over related work and include different resilience issues in the controller placement process. We argued that it is not sufficient to look only at node-to-controller latencies but a controller placement should also fulfill certain resilience constraints especially for the control plane.

Characterization of BitTorrent swarms and their distribution in the Internet

The optimization of overlay traffic resulting from applications such as BitTorrent is a challenge addressed by several recent research initiatives. However, the assessment of such optimization techniques and their performance in the real Internet remains difficult. Despite a considerable set of works measuring real-life BitTorrent swarms, several characteristics of those swarms relevant for the optimization of overlay traffic have not yet been investigated. In this work, we address this lack of realistic swarm statistics by presenting our measurement results. In particular, we provide a statistical characterization of the swarm sizes, the distribution of peers over autonomous systems (ASs), the fraction of peers in the largest AS, and the size of the shared files. To this end, we consider different types of shared content and identify particular characteristics of regional swarms. The selection of the presented data is inspired by ongoing discussions in the IETF working group on application layer traffic optimization (ALTO). Our study is intended to provide input for the design and the assessment of ALTO solutions for BitTorrent, but the applicability of the results is not limited to that purpose.

Using buffered playtime for QoE-oriented resource management of YouTube video streaming

YouTube is the most important online platform for streaming video clips. The popularity and the continuously increasing number of users pose new challenges for Internet service providers. In particular, in access networks where the transmission resources are limited and the providers are interested in reducing their operational expenditure, it is worth to efficiently optimise the network for popular services such as YouTube. In this paper, we propose different resource management mechanisms to improve the quality of experience (QoE) of YouTube users. In particular, we investigate the benefit of cross-layer resource management actions at the client and in the access network for YouTube video streaming. The proposed algorithms are evaluated in a wireless mesh testbed. The results show how to improve the YouTube QoE for the users with the help of client-based or network-based control actions. Copyright

Modeling and experimenting combined smart sleep and power scaling algorithms in energy-aware data center networks

Abstract Recent surveys have shown that the energy consumption in a data center considerably contributes to its operation costs. A remarkable part of the large energy volume consumed in data centers today is due to the over-provisioning of such network resources as switches, links, and servers to meet the stringent requirements on reliability. Therefore performance and energy issues are important factors for the design of large multi-tier data centers that can support multiple services. However, the design, analysis, and experiments of such a large and complex system often suffer from the lack of appropriate experimental infrastructures. In this paper, we firstly propose a new energy saving scheme that combines smart sleeping and power scaling algorithms. An energy analysis model is then proposed to calculate the energy saving bounds in case of low and high traffic utilization. We also present a platform for in-depth analysis of energy-aware data center networks, which is a combination of hardware testbed and emulation. Based on OpenFlow technology, the experimental platform is designed to capture details of energy consumed by all network components such as links, ports, and switches under different scenarios. Analytical and emulation results show that the combined algorithm improves the energy saving under the varied traffic utilization.

Optimizing unique shortest paths for resilient routing and fast reroute in IP-based networks

Intradomain routing in IP networks follows shortest paths according to administrative link costs. When several equal-cost shortest paths exist, routers that use equal-cost multipath (ECMP) distribute the traffic over all of them. To produce single-shortest path (SSP) routing, a selection mechanism (tie-breaker) chooses just one of the equal-cost paths. Tie-breakers are poorly standardized and use information that may change over time, which makes SSP routing unpredictable. Therefore, link costs producing unique shortest paths (USP) are preferred.

Demonstrating the optimal placement of virtualized cellular network functions in case of large crowd events

This demonstration shows how Network Function Virtualisation (NFV) [1] can be used by a network provider to dynamically provide required mobile core network functions in case of a large ”Mega” event like a soccer game or a music festival. Economical reasons may not justify the installation or continuous maintenance of expensive dedicated hardware which is necessary to cope with the high load generated by visitors of such an event only in some parts of the network and only for a short time. The Evolved Packet Core (EPC) in nowadays’ mobile LTE networks consists of several, specialized components: first pure control-plane elements like Mobility Management Entities (MME) which can be installed on virtualized IT hardware in the cloud already today and second gateways which are a mixture of controland user-plane. This demonstration focuses on the Serving Gateway (SGW) which switches GTP (GPRS Tunneling Protocol) tunnels in a LTE network. In such an LTE network with ≈ 10 mio. subscribers about 10 SGW devices are in use. On the way to a successful deployment of NFV-based EPC components, several challenges have to be met. This includes the deployment, interconnection and configuration of LTE components in the cloud. As such, entities that instantiate and orchestrate the virtualized functions are required. The presented demonstration shows a scenario of NFVbased dynamic capacity addition to a LTE mobile network, indicated in Figure 1. By incorporating the new demands added by the increased access capacity, virtualized SGW instances are launched at the optimal locations in the network, by re-programming SDN enabled network elements (NE+). Figure 1(a) shows the normal configuration of the LTE network including the statically located SGW devices as well as

POCO-PLC: Enabling dynamic pareto-optimal resilient controller placement in SDN networks

Recently, Software Defined Networking (SDN) has gained a lot of attention. The paradigm shift towards centralized architectures with a separation of control plane and data plane is expected for several use cases, including amongst others core communication networks, data center networks, or Network Function Virtualization (NFV). An important issue during SDN deployment is the placement of controllers in the network. The first to address this topic were Heller et al. [3] followed by other researchers who added different resilience issues into their observations. An overview on relevant publications can be found in our previous work [1], where we also introduced our POCO toolset. Here, a powerful and user friendly POCO-PLC toolset is provided and presented that facilitates the analysis and optimization of the controller placement in SDN networks under dynamic conditions. A general overview on the functionality and architecture of POCO-PLC is given. Then, possible use cases enabled by POCO-PLC is indicated.

QoS provisioning in WLAN mesh networks using dynamic bandwidth control

WLAN, based on the IEEE 802.11 standard has been extensively studied since its release. In addition to infrastructure access to WLAN, mesh networks currently attract a lot of attention. This comes from the envisioned advantages of wireless mesh networks, such as cheap installation costs, extended coverage, robustness, easy maintenance, and self-configuration possibilities. In this paper we focus on Quality of Service support for multimedia applications in WLAN-based mesh networks. Therefore, a dynamic bandwidth control mechanism is implemented on the network layer and the results show that high prioritized traffic can be protected from disturbing best effort traffic.

Network configuration with quality of service abstractions for SDN and legacy networks

In this paper, we demonstrate an implementation of a Network Services Abstraction Layer (NSAL) on top of the network control and management plane. Furthermore, we introduce a unified data model for both Software Defined Networking (SDN) and legacy devices that allows managing and configuring both networks in a unified way in order to achieve Quality of Service (QoS) for time-critical tasks (e.g. VoIP). Due to the unified data model, network operators are able to manage their network through one interface. We demonstrate a use case by implementing a VoIP scheduling application on top of the NSAL and evaluate VoIP call quality in a distributed heterogeneous network.