Software Switch Research Articles

Toward Greater Intelligence in Route Planning: A Graph-Aware Deep Learning Approach

Software-defined networking decouples the control plane and data plane, which grants more computing power for routing computations. Traditional routing methods suffer from the complex dynamics in networking, and they are facing issues such as slow convergence and performance decline. Deep learning techniques have shown preliminary results on solving the routing problem, and bringing more accuracy, precision, and intelligence compared with traditional modeling techniques. However, the existing deep learning architectures are not built to learn from the crucial topological relations between forwarding nodes, which restricts the model’s ability to handle different network conditions. In this paper, we propose a deep learning based intelligent routing strategy with revised graph-aware neural networks, which learns topological information efficiently. In addition, a set of features suitable for network routing are designed so that the networking state are well represented upon each routing decision. In experiments, the performance of the proposed work is demonstrated with a real-world topology and the production level software switches. The execution time is evaluated on various kinds of network topology and different network scales. Also, the simulation result shows that the proposed work is more accurate and efficient compared to the state-of-the-art routing strategy.

ARP-P4: deep analysis of a hybrid SDN ARP-Path/P4Runtime switch

The Software-Defined Networking (SDN) architecture decouples the control plane from the data plane, but it does not explicitly state where the control should be located. This article analyses the benefits of maintaining the control as close as possible to the data plane, instead of the more traditional centralised control plane approach. To this purpose, it delves into the study of ARP-P4, a hybrid software switch defined by using the P4 language to facilitate its future use and deployment in P4 targets. Its hybrid properties come from supporting two complementary different ways of establishing paths: a centralised SDN approach based on P4-Runtime and a traditional distributed approach based on the ARP-Path protocol that obtains a similar performance to centralised solutions based on Equal Cost Multi-Path (ECMP) and Dijkstra. The results show the feasibility of hybrid devices that combine different forwarding paradigms without losing performance with respect to well-known solutions such as ECMP, and how their combined use can lead to enhance and scale communication networks.

Analytical Modelling of Software and Hardware Switches with Internal Buffer in Software-Defined Networks

OpenFlow supports internal buffering of data packets in Software-Defined Networking (SDN) switch whereby a fraction of data packet header is sent to the controller instead of an entire data packet. This internal buffering increases the robustness and the utilization of the link between SDN switches and the controller by absorbing temporary burst of packets which may overwhelm the controller. Existing queuing models for an SDN have focused on the switches that immediately sends packets to the controller for decisioning, with no existing models investigating the impact of the internal buffer in SDN software and hardware switches. In this paper, we propose a unified queueing model to characterise the performance of SDN software and hardware switches with the internal buffer. This unified queueing model is an analytical tool for network engineers to predict a delay and loss during SDN deployments in delay and loss sensitive environments. Our results show that a hardware switch achieves up to 80% lower average packet transfer delay and 99% lower packet loss rate at the cost of requiring up to 50% more queue capacity than a software switch. The proposed models are validated with a discrete event simulation, where the error between 0.6% and 2.8% was observed for both average packet transfer delay and average packet loss rate. Moreover, a hardware switch outperforms a software switch with increasing number of hosts per switch suggesting that a hardware switch has better scalability. We use the insights from the model to develop guidelines that help network engineers decide between a software and hardware switch in their SDN deployments.

Software switches: novel hands-free interaction techniques for quadriplegics based on respiration–machine interaction

The majority of interaction techniques for individuals with disabilities to control a computer require an additional dedicated hardware like switches beyond standard computer peripherals. Furthermore, each traditional hardware switch emulates different keyboard characters or mouse clicks depending on its manufacturer. There is no commonly agreed standard on this; while some switch accessible programs expect to receive enter character, other programs expect to receive mouse left click. Switch manufacturers, therefore, distribute software drivers, which allow users to assign expected characters, with switches to overcome this standardization problem by making them adaptable. As software drivers are developed for a specific switch, they are not compatible with the other manufacturers’ switches. On the other hand, for the ones with a very limited motor activity but a complete respiration activity, sip-and-puff switches are efficient solutions. However, they are expensive and invasive systems with tubes, that have to be changed regularly due to the hygiene concern, inside users’ mouth. Although invasive respiration-based systems (sip-and-puff devices) caught enough attention by researchers over the years, using non-invasive respiration-based interaction techniques is an underestimated approach in general. In this study, we proposed two novel non-invasive interaction techniques as software switches (PuffCam and PuffMic) compatible with any switch accessible software. Both techniques are respiration operated, where a hard puff, detected by a microphone or an adapted camera, is considered as ‘switch-on’ like a puff switch. We have also collected the objective data by an evaluation software namely TestBed. A user study (conducted with 46 participants with/out disabilities) revealed that the accuracy, precision, recall, and false positive rate of our interaction techniques were quite impressive, and PuffCam performed better than PuffMic for all metrics. According to questionnaire findings, comfort assessment of interaction techniques by participants was rated quite satisfactory. All participants agreed that the idea of controlling a computer via breathing without purchasing any dedicated hardware sounded very promising. Because most interaction techniques for computer control require extra dedicated devices and there is not any adaptable (i.e., compatible with most switch accessible software) software switch to the best of our knowledge, proposed interaction techniques can help community in an open access manner without purchasing any device.

A novel software-defined network packet security tunnel forwarding mechanism.

The OpenFlow protocol match field capacity is fixed and limited, and packet forwarding in software-defined network lacks valid authentication of data source, integrity verification, and confidentiality protection mechanism. OpenFlow only supports the MPLS label tunnel establishment, and therefore cannot establish a secure tunnel flexibly. In order to solve these problems, we propose P4Sec, a novel software-defined network packet security tunnel forwarding mechanism. As P4 allows the data plane to be reprogrammed to realize the characteristics of packet forwarding, we build a software-defined network security tunnel to prevent data malicious tampering, stealing, forgery and other malicious network behavior, implementing packet routing and forwarding based on gateway identity. Finally, we construct a P4Sec prototype system based on the software switch BMv2, verify the effectiveness of the mechanism through experimental analysis, and evaluate the overhead of the mechanism. The results demonstrate that P4Sec security mechanism ensure the authenticity, integrity, and confidentiality of forwarded data, and realize the secure forwarding requirements of data packets in software-defined network.

WhiteRabbit: Scalable Software-Defined Network Data-Plane Verification Method Through Time Scheduling

Software-defined networks are vulnerable to attacks by compromised switches because commonly used programmable software switches are risky than traditional hardware ones. Although several countermeasures have been proposed to address compromised switches, the accuracy of detecting a malicious behavior depends on the performance of network statistics gathering by a controller. In this paper, we propose that WhiteRabbit is an approach to verify the consistency of the forwarding state by gathering real-time network statistics gathering from switches with accurate time scheduling. WhiteRabbit can detect attacks by compromised switches without being influenced by the performance of statistics gathering of a controller. Given that the proposed utilizes moving average, it mitigates the effect on the verification accuracy from the impact of the switch performance, such as scheduling error. In our previous work, we demonstrated the feasibility of WhiteRabbit using a prototype system. However, we could not evaluate the impact of the difference between the scheduled and actual execution times in our previous work, because we performed the experiment in a minimal setup using Mininet. Thus, we measured the scheduling error and time required to gather statistics in a large-scale environment. We also confirmed that the scheduling error is lower than the time required to gather statistics. Additionally, considering that WhiteRabbit only depends on the scheduling error, we verified that the accuracy of WhiteRabbit is higher than prior arts on the tree topology constructed with 15 switches.

ZCopy-Vhost: Replacing Data Copy With Page Remapping in Virtual Packet I/O

Virtualization technology is the core technology of cloud computing. While virtualization technology offers flexibility in many ways, it also introduces additional performance overhead. For network systems, it needs to provide additional virtual switching, the virtual packet I/O, and other functions. Providing these features requires a lot of CPU resources. As cloud services grow, more and more CPU resources are being used to provide virtualization support, which means that fewer CPUs can be sold to tenants. Therefore, cloud service providers have been constantly seeking more efficient virtualization solutions for network systems. In this paper, we identify one of the main reasons for the large consumption of CPU resources in the virtual networking system - copy and propose to use memory page remapping technology to eliminate the copy. Moreover, to adapt to the zero-copy technology, we have integrated the technology with a mature virtual switch software in the host and a userspace protocol stack in the virtual machine. These components together build a more efficient network system in virtual environments. The evaluations with microbenchmarks and macrobenchmark show that our system performs much better than the state-of-the-art solutions.

Software-Defined MANET Swarm for Mobile Monitoring in Hydropower Plants

The construction and running of hydropower plants does not merely involve hydropower generation itself but also facilities surveillance, water quality monitoring, harmful creatures tracking, etc., thus requiring mobile and collaborative monitoring capabilities. MANET (Mobile Ad Hoc Networks), with its mobility, flexibility, and robustness in volatile networking environment, is a competitive candidate to fulfill such tasks. Nevertheless its distributed structure prevents the effective collaboration between MANET nodes. SDN (Software-Defined Networking) provides the centralized control over the network underlay. This paper proposes the SDN-controlled IEEE 802.11 MANET swarm for mobile monitoring in hydropower plants. First, the MANET node is implemented by integrating Raspberry Pi with cameras, various sensors, etc., into the low-cost wheeled mobile hardware to enable sensibility and mobility. Then, multiple such MANET nodes are networked through ad hoc protocols, to construct a flexible and distributed MANET underlay; meanwhile, to implement centralized control over the MANET underlay, every MANET node is equipped with OpenFlow switch software (e.g., Open vSwitch) so that OpenFlow directives issued by the SDN controller can be understood and executed, hence the SDN overlay on top of the MANET underlay. Finally, OpenFlow is extended to offer physical actions such as mobility and sensibility manipulation, beyond pure data forwarding in traditional SDN applications, to achieve mobile monitoring in hydropower plants. The SDN-controlled MANET swarm features mobile and sensing capabilities, flexible networking through ad hoc protocols, and efficient and unified control by SDN. Experiment results prove the feasibility of this network architecture.

Flow Setup Latency in SDN Networks

In software-defined networking, the typical switch-controller cycle, from generating a network event notification at the controller until the flow rules are installed at the switches, is not an instantaneous activity. Our measurement results show that this has serious implications on the performance of flow setup procedure, specifically for larger networks: we observe that, even with software switches, the flow setup latency for networks of around 500 switches is in the order of 50 ms, with 99th percentile exhibiting $10\times $ higher latencies. To reduce both the latency and the variance of the flow setup, we propose path aggregation strategies, which turn the network into a set of pre-configured pipes that connect any pair of nodes. Our approach radically simplifies the flow setup procedure by minimizing the set of switches to be updated for new user-initiated flows to a constant number. We implement our solution in our testbed and study its performance through measurements. The results show that in similar settings, it reduces the median and 99-percentile latencies to 5.9 and 7 ms, respectively, significantly improving the performance, especially in the tail.

P4 Edge Node Enabling Stateful Traffic Engineering and Cyber Security

Next-generation edge nodes interfacing innovative IT clusters, 5G fronthaul, and internet of things (IoT) gateways to the optical metro/core network will require advanced and dynamic online quality of service (QoS) per-flow traffic treatment, assuring ultra-low latency requirements. However, current software-defined networking (SDN) implementations (e.g., OpenFlow) do not support forwarding procedures based on the network state, profile variations, and the history of flow statistics at the node level. Currently, such procedures require intervention by the SDN controller, leading to scalability issues and additional latency in data plane forwarding. Moreover, severe security challenges are expected to affect such nodes and threaten IT resources. Thus, increasing bandwidths will require direct deep packet inspection to avoid involvement of the SDN controller, as performed currently, or dedicated and costly security systems. This paper leverages on the potential of the programming protocol-independent packet processors (P4) open source language, recently introduced by the inventors of OpenFlow, to program the data plane structure and behavior of an SDN switch. P4 is able to instantiate custom pipelines and stateful objects, enabling complex workflows, user-defined protocols/headers, and finite state machines enforcement. Moreover, P4 allows portable implementations over different hardware targets, thus opening the way to open source fully programmable devices. Special effort is dedicated to motivate and apply P4 within a multilayer edge scenario, proposing the architecture and the applicability of an SDN P4-enabled packet-over-optical node. Moreover, three specific multilayer use cases covering dynamic traffic engineering (TE) (e.g., traffic offload and optical bypass) and cybersecurity (e.g., distributed denial of service port scan) are discussed and addressed through P4-based solutions. Experimental evaluations have been conducted over a multilayer SDN network exploiting reference P4 software switches (i.e., the behavioral model version 2, or BMV2) and field-programmable gate arrays (FPGAs) at 10 gigabit Ethernet optical interfaces. Extensive results report effective dynamic TE and cybersecurity mitigation enforcement at P4 switches without any controller intervention, showing excellent scalability performance and overall latencies practically in line with current commercial OpenFlow switches.

Modelling Software-Defined Networking: Software and hardware switches

In Software-Defined Networking (SDN), a switch is a forwarding device that moves data packets in a network. A software switch handles forwarding functions in software and thus cannot forward packet at line speed while a hardware switch leverages optimised forwarding hardware to forward packets at line speed. However, there has been very little research in the literature to help network engineers understand the tradeoffs in choosing one over the other. In this paper, we develop a unified queueing model for characterizing the performance of hardware switches and software switches in SDN. The unified queueing model is an analytical tool for engineers to predict delay, packet loss and throughput in their SDN deployments. Existing queueing models of SDN have focused on performance analysis of software switches, while our work presented herein is the first to present a unified analysis of hardware and software switches. Our proposed models exhibit errors below 5% compared to simulation. Between a hardware and software switch, the evaluation shows that a hardware switch achieves an average 80% lower delay and up to 100% lower packet loss probability compared to a software switch. The more a hardware switch involves the controller for decisioning, the lower the gains in terms of packet delays through the switch.

Evaluation of video payload over low latency networks: Flexilink

ABSTRACTBroadcast organisations need to handle both time deterministic traffic and best effort traffic. The adoption of packet networks allows a single converged network for handling both real-time (RT) media traffic as well as data traffic. Due to the historic pervasive growth of IP network, they are widely used to realise this converged network. However, they do have several drawbacks like big header size, software switching delays and inability to support multiple addressing schemes. To overcome these issues, solutions like traffic engineering, quality of service and over-provisioning are implemented in IP networks but has resulted in higher cost, power and system requirements. Flexilink network has been tested for audio payloads using simulation model and outperforms any existing RT network by providing an exceptional average end-to-end delay of microsecond range and a jitter below the audible threshold for live digital audio streaming. However, it is not tested for video payloads. A review of various networks used RT media distribution is included in this paper. The Flexilink architecture design modification to support the most widely used video formats such as MPEG-TS, MPEG-PS and SDI streams is explained. The average delay for multiple video streams over Flexilink is provided using simulation model. The recommendation for testing Flexilink using hardware is also explained in this paper.The Flexilink Mac design uses dual-buffer system to handle AF (asynchronous flow) and SF (synchronous flow) separately, with the Frame structure that supports current Ethernet frames.

Dynamic Multi-level Auto-scaling Rules for Containerized Applications

Container-based cloud applications require sophisticated auto-scaling methods in order to operate under different workload conditions. The choice of an auto-scaling method may significantly affect important service quality parameters, such as response time and resource utilization. Current container orchestration systems such as Kubernetes and cloud providers such as Amazon EC2 employ auto-scaling rules with static thresholds and rely mainly on infrastructure-related monitoring data, such as CPU and memory utilization. This paper presents a new dynamic multi-level (DM) auto-scaling method with dynamically changing thresholds, which uses not only infrastructure, but also application-level monitoring data. The new method is compared with seven existing auto-scaling methods in different synthetic and real-world workload scenarios. Based on experimental results, all eight auto-scaling methods are compared according to the response time and the number of instantiated containers. The results show that the proposed DM method has better overall performance under varied amount of workloads than the other auto-scaling methods. Due to satisfactory results, the proposed DM method is implemented in the SWITCH software engineering system for time-critical cloud applications.

FAS: Using FPGA to Accelerate and Secure SDN Software Switches

Software-Defined Networking (SDN) promises the vision of more flexible and manageable networks but requires certain level of programmability in the data plane to accommodate different forwarding abstractions. SDN software switches running on commodity multicore platforms are programmable and are with low deployment cost. However, the performance of SDN software switches is not satisfactory due to the complex forwarding operations on packets. Moreover, this may hinder the performance of real-time security on software switch. In this paper, we analyze the forwarding procedure and identify the performance bottleneck of SDN software switches. An FPGA-based mechanism for accelerating and securing SDN switches, named FAS (FPGA-Accelerated SDN software switch), is proposed to take advantage of the reconfigurability and high-performance advantages of FPGA. FAS improves the performance as well as the capacity against malicious traffic attacks of SDN software switches by offloading some functional modules. We validate FAS on an FPGA-based network processing platform. Experiment results demonstrate that the forwarding rate of FAS can be 44% higher than the original SDN software switch. In addition, FAS provides new opportunity to enhance the security of SDN software switches by allowing the deployment of bump-in-the-wire security modules (such as packet detectors and filters) in FPGA.

VirtPhy: Fully Programmable NFV Orchestration Architecture for Edge Data Centers

Emerging paradigms, such as edge computing, require the geographical distribution of small-scale data centers that will use network functions virtualization (NFV) to provide new services with stringent demands related to throughput, latency, cost, innovation, and efficient orchestration. To tackle these demands, this paper proposes VirtPhy, a fully programmable architecture for NFV orchestration in edge data centers, based on server-centric topologies, software-defined networking, software switches, distributed service chaining, and source routing. A proof-of-concept is implemented in a data center testbed using OpenStack cloud platform in a hypercube topology. The results show its orchestration mechanisms can efficiently provision NFV service requests by steering traffic through a sequence of virtualized network functions in a server-centric topology.

To Verification Infinity and Beyond

When I worked for ATT someone made a tiny change to telephone switch software, the change was thought too small to be worth verifying, and due to some obscure interactions between telephone switches, much of the East Coast of the United States lost service.

The P416 Programming Language

P4 is a language for expressing how packets are processed by the data-plane of a programmable network element such as a hardware or software switch, network interface card, router or network function appliance. This document describes the most recent version of the language, P416, and the reference implementation of the P416 compiler.

Wireless Transport Emulator Based on LINC Software Switch

Recent technological advances in the telecommunications industry have led to discovering the limitations of traditional communication networks and, in this context, to the emergence of a new paradigm that tries to mitigate them, namely software defined networking (SDN). SDN is beginning to infiltrate all the aspects of a network, and all kind of networks, from LANs, campus, enterprise, or mobile operator networks. Since this field is not mature yet, the interest from the academia and industry is very strong, thus having the right tools for research is critical. In this respect, mininet is a very good simulator of software defined networks. A software switch that can emulate optical extensions of SDN has emerged and has been integrated with mininet: LINC-OE. There is a need in the research community for a software switch that can be integrated with mininet and used to emulate wireless transport networks. This paper proposes a solution for that need by extending the LINC-OE for wireless transport: LINC-WE. In this context, two approaches are considered: the first is emulating the wireless links between two nodes and the other is integrating the LINC switch with a NETCONF server, for being able to expose a wireless transport device information model for simulations.

Optimized Three-Stage Strategy of Orderly Management of File Stream in Network Printing Workflow

Abstract This paper proposed the optimized three-stage strategy of orderly management of file stream in network printing workflow. In first stage the customer preflight was added to the conventional prepress preflight, which set quick-type, general-type, stable-type and comprehensive-type preflight scheme for customers by investigating the possibility of occurrence of errors of various preflight entries and customer relevancy. Based on four types of preflight schemes, four customer preflight applications were created by using Pitstop Pro and Connector You. Customers can select one of the appropriate applications for quick preflight before delivering to reduce the probability of returning the files back for correction when errors occur in preflight. Second stage is printing enterprise prepress preflight, which proposed the concept of the prepress preflight tolerance. In addition, according to the different requirements of printing technology and quality, this paper established the prepress preflight scheme based on the printing enterprises’ tolerance for preflight error. Third stage is to manage the optimized sorting of prepress file stream. To achieve it, the SPT rule and the 0-1 programming model were adopted, time spend of preflight, printing process and post-press finishing was also taken into account. Moreover, Enfocus Switch software was applied to establish automated network printing workflow and complete the file stream sorting. Finally, the result of simulation analysis of the optimized strategy shows that the optimized three-stage strategy can effectively shorten the file processing time.

A Lossless Network for Data Acquisition

The bursty many-to-one communication pattern, typical for data acquisition systems, is particularly demanding for commodity TCP/IP and Ethernet technologies. We expand the study of lossless switching in software running on commercial off-the-shelf servers, using the ATLAS experiment as a case study. In this paper, we extend the popular software switch, Open vSwitch, with a dedicated, throughput-oriented buffering mechanism for data acquisition. We compare the performance under heavy congestion on typical Ethernet switches to a commodity server acting as a switch. Our results indicate that software switches with large buffers perform significantly better. Next, we evaluate the scalability of the system when building a larger topology of interconnected software switches, exploiting the integration with software-defined networking technologies. We build an IP-only leaf-spine network consisting of eight software switches running on distinct physical servers as a demonstrator.