In modern dynamic and heterogeneous network environments, the limitations of traditional routing algorithms have become increasingly apparent. While these classical methods have provided robust solutions over decades, they fall short in meeting the current demands for scalability, adaptability, and simplicity. In contrast, Layer-2 networking offers a decentralized, plug-and-play paradigm that is well suited for rapid deployment across diverse scenarios, including data center networks or softwarized services for edge computing networks. However, existing solutions suffer from prolonged convergence times and high control overhead, which hampers their efficiency in dynamic topologies and deviates from the autoconfiguration and plug-and- play essence of a pure Layer-2 approach. In this paper, we present BareFlow, a novel pure switching mechanism that obviates the need for explicit control messages by operating without modifying packet headers. We evaluated BareFlow using the NS-3 simulator in a Spine-Leaf topology (usual in data center networks) under various traffic distributions. These preliminary results demonstrate that BareFlow achieves near-maximum throughput for large flows and efficiently manages flow completion times across different flow sizes. These findings highlight the potential of BareFlow as a scalable, self-configuring solution, offering efficient and adaptive forwarding with significantly reduced overhead, which is particularly suited for edge computing environments, in which devices are usually constrained in capacity.

Emerging applications such as telemedicine, the tactile Internet or live streaming place high demands on low latency to ensure a satisfactory Quality of Experience (QoE). In these scenarios the use of trees can be particularly interesting to efficiently deliver traffic to groups of users because they further enhance network performance by providing redundancy and fault tolerance, ensuring service continuity when network failure or congestion scenarios occur. Furthermore, if trees are isolated from each other (they do not share common communication elements as links and/or nodes), their benefits are further enhanced since events such as failures or congestion in one tree do not affect others. However, the challenge of computing fully disjoint trees (both link- and node-disjoint) introduces significant mathematical complexity, resulting in longer computation times, which negatively impacts latency-sensitive applications. In this article, we propose a novel algorithm designed to rapidly compute multiple fully (either link- or node-) disjoint trees while maintaining efficiency and scalability, specifically focused on targeting the lowlatency requirements of emerging services and applications. The proposed algorithm addresses the complexity of ensuring disjointedness between trees without sacrificing performance. Our solution has been tested in a variety of network environments, including both wired and wireless scenarios. The results showcase that our proposed method is approximately 100 times faster than existing techniques, while achieving a comparable success rate in terms of number of obtained disjoint trees. This significant improvement in computational speed makes our approach highly suitable for the low-latency requirements of next-generation networks.

The incorporation of end devices in the edge-to-cloud continuum yields substantial benefits to conventional cloud computing frameworks, expediting communication between end devices and computational resources, and resulting in new use cases, particularly in the field of mobile networks. However, few related works leverage the full potential of resource sharing at the far edge, as most proposals require that end nodes rely on a higher-capacity computational node. This manuscript presents Multi-Hop Wireless Resource Sharing Protocol (MuHoW), a lightweight protocol tailored for multi-hop wireless networks. MuHoW enables resource sharing within collaborative edge-computing networks by facilitating the discovery of wireless neighbours and subsequently establishing a confluence tree directed towards the edge/fog infrastructure. This tree serves as a conduit for aggregating essential resource sharing information, ensuring the establishment of a seamless collaborative edge/fog environment. The empirical findings highlight the scalability of MuHoW, due to its linear control message growth with network size. Moreover, the efficiency of the protocol is very high even in lossy environments as evidenced by the fact that most resource sharing messages are successfully delivered as expected.

Network softwarization has significantly evolved since programmable data planes became topical in academia and industry. Programming Protocol-Independent Packet Processors (P4) is a language to define packet forwarding behavior. Forwarding devices that are programmed with the P4 language support a flexible way to define headers, parse graphs, and data plane logic. However, extending the data plane with additional functionalities has an impact on packet data plane latency. For this reason, this paper analyzes the key factors that affect data pane latency to packets processed by the Tofino-based target (Tofino Native Architecture (TNA)), which can be considered the de facto production-ready and P4-programmable Application-Specific Integrated Circuit (ASIC). Our work first provides an extensive set of latency measurements and, afterwards, it includes a set of data plane latency predictions using the model derived from the latency results and machine learning (ML) algorithms. We demonstrate that the PCA-lasso polynomial (PLP) obtains the best results among the algorithms tested. The best-case results show that PLP obtained an accuracy of 98.22% prediction accuracy when considering the parser, deparser, and the control block for traffic running at 10G/s (SFP+) and 100G/s (QSFP28). To the best of our knowledge, this is the first work that provides such a comprehensive profiling, including a method to predict data plane latency in production-grade Tofino ASIC-based switching hardware, which could be leveraged to yield accurate latency values prior to investment and deployment.

RPL is the de-facto IPv6-based routing protocol for the Internet of Things (IoT), where networks are mainly formed by sensors and other low capacity devices. However, RPL lacks scalability and is inefficient in any-to-any communications. In this article, we present IoTorii, a hierarchical protocol that creates routes based on a probe frame, instead of the computation of a distance vector, as in RPL. We implemented a proof-of-concept of IoTorii and compared it with RPL, proving lower number of table entries, decreased exchanged control messages and faster convergence time, with similar path hop count, which is particularly promising in the IoT field.

Rapid Spanning Tree Protocol (RSTP) is a layer 2 protocol that blocks redundant links to avoid loops. Due to its well-known limitations, diverse solutions have been proposed to replace it, but scarcely adopted, such as Transparent Interconnection of Lots of Links (TRILL) and Shortest Path Bridging (SPB). To overcome the constraints of RSTP, while keeping its advantages, we propose to combine the fast and proven reconfiguration mechanisms of RSTP with the shortest path capabilities and load balancing of ARP-Path bridging in the same device. The combined bridges are self-configuring and allow full miscibility of combined bridges in legacy RSTP networks. This enables a smooth migration path from standard 802.1D bridges to ARP-Path shortest path bridges. This paper presents a preliminary proposal: full validation, simulation, implementation and evaluation are ongoing.

This paper presents ARP-P4, a hybrid switch based on a local ARP-Path control plane written in P4, but maintaining legacy P4 Runtime control plane capabilities. Its main purpose is to study the readiness of P4 to perform directly local control actions at the data plane, such as the required ones for the ARP-Path protocol. Thus, ARP-P4 defines a data plane that forwards any ingress packet locally via ARP-Path or remotely via P4 Runtime installed rules through an SDN controller. Finally, we discuss the current limitations of P4 non-standard externs to interact with P4 Runtime to develop autonomous capabilities.

Scaling Ethernet switched networks to sizes of hundreds of thousands of hosts is a key requirement for data center networks. PAST is a simple and scalable OpenFlow-based layer-two architecture for data center Ethernet networks that outperforms Equal Cost Multipath Forwarding. PAST computes and configures one spanning tree for every MAC destination address. However, an important bottleneck of PAST is the time taken to perform the massive installation of rules at OpenFlow switches to set up the trees. PAST-HOAS aims to solve that obstacle, building trees instantly by path discovery, either triggered by the controller with a multicast probe frame from the edge destination switch or by the standard ARP Request frame; without flow installation and without tree computation. PASTHOAS is loop-free, and provides 100-400 times faster network initialization, lower forwarding delays and increased robustness and scalability.

Routing via multiple paths offers many advantages such as traffic separation, congestion reduction and quality of service policies in enterprise, campus and data center networks. Disjointness of the multiple paths may provide extra benefits like increased reliability and improved traffic separation. We propose DMDP, a layer two protocol that discovers multiple disjoint paths in a switched network by exploration of the network with multicast frames. DMDP provides fast discovery of node-disjoint or link-disjoint paths, selecting the fastest disjoint paths available without any route computation over the network graph. Paths obtained by DMDP are similar to the multiple paths computed with a k-shortest path algorithm, but much simpler to find. Multiple disjoint path search may be implemented as a proactive, reactive on-demand or hybrid service at edge switches.

Software-Defined Networks (SDN) are constantly evolving and so is their software. One of the key advantages of SDN over traditional networks is the ability to rapidly develop and deploy new features. However, updating them often requires restarting the SDN controller and causes network downtime. In addition to such planned updates, unforeseen, accidental downtime is also a risk for SDN networks. While commercialized SDN controllers are adding mechanisms to deal with both planned and accidental downtime, they still are not competitive with conventional approaches, which typically use redundant hardware and special software to address these problems. In this paper we investigate how these two challenges to SDN can be addressed with dynamic reconfiguration and show how the state of the network can be managed by reconfiguration. Finally, we present a proof-of-concept implementation of our approach.

The Software-Defined Networking (SDN) ecosystem is still characterized by a multitude of different controller platforms, each with its own programming model, execution model, and capabilities. This creates a danger of a controller lock-in for both developers of SDN control applications and operators of SDN networks. Since no single controller platform appears to dominate the ecosystem for the foreseeable future, there is a need for portability of control applications between different platforms. We propose an architecture based on executing multiple instances of different controller platforms concurrently in a network to provide the SDN code the environment it was written for. It is built around a controller-independent network event routing element called Network Engine that provides composition and conflict resolution. Results obtained in realistic scenarios demonstrate the feasibility of the proposed approach, which increases both developer productivity and operational flexibility. A preliminary prototype of the architecture is available for testing as an open source project.

The Software-Defined Networking (SDN) paradigm allows networking hardware to be made “malleable” and remotely manageable by the so-called SDN controllers. However, the current SDN landscape is extremely fragmented. Different open and closed source controller frameworks such as Open-Daylight [1], Ryu [2], Floodlight [3], etc. exist. Porting SDN applications from one such platform to another is practically impossible and so, SDN users like network operators face a situation where they are either confined to applications working for the platform of their choice, or forced to re-implement their solutions every time they encounter a new platform.

ARP-Path is a simple, low latency, shortest path bridging protocol for campus, enterprise and data center networks. We recently found that this protocol natively distributes the traffic load in networks having redundant paths of similar characteristics. The reason is that every new path between hosts is selected on-demand in a race among ARP Request packet replicas over all available paths: the first arriving replica gets its path selected on the fly. This means a continuous adaptation of new paths to variations on the load at links and bridges. To show this unique load distribution capability and path diversity property we use a number of simulations for complex scenarios, including two different simulators: one flow-based and one packet-based, and two basic topologies: data center and a regular mesh. We also verify this behavior on real hardware on a network of nine ARP-Path NetFPGA switches. The conclusion is that the ARP-Path protocol efficiently distributes traffic via alternative paths at all load levels, provided that multiple paths of similar propagation delays are available.

This paper describes Torii-HLMAC, a scalable, fault-tolerant, zero-configuration data center network fabric architecture protocol as a fully distributed alternative to PortLand for similar multiple tree network topologies. It uses multiple, fixed, tree-based positional MAC addresses for multiple path and table-free forwarding. Addresses are assigned by a simple extension of the Rapid Spanning Tree Protocol. Torii-HLMAC enhances the PortLand protocol advantages of scalability, zero configuration and high performance and adds instant path recovery, fully distributed routing and address assignment. ARP Proxy may be used to avoid ARP broadcast messages.

This paper describes Flow-Path, an AllPath flow-based switching protocol that features improved load adaptive properties. Upon arrival of a new flow to the network, it explores every possible path reaching from source to destination host and selects the lowest latency path at the moment. It is based on the same basic principle than ARP-Path, that is, snooping the ARP protocol dialog (request and reply messages) to explore all available paths at the same time that address resolution takes place, but it is flow-based instead of source address-based. While preserving the main advantages of ARP-Path: shortest path bridging exploiting the full network topology, Flow-Path has the advantages of full independence of flows at the time of path creation and guarantees path symmetry (congruency) and increased path diversity. Flow-Path thus improves load distribution, at the expense of increased address table size in each bridge.

This paper describes the implementation of ARP-Path (a.k.a. FastPath) bridges, a recently proposed concept for low latency bridges, in Linux/Soekris and OpenFlow/NetFPGA platforms. These ARP-based Ethernet Switches rely on the race between the replicas of a standard ARP Request packet flooded over all links, to discover the minimum latency path to the destination host, complemented in the opposite direction by the ARP Reply packet directed to the source host. Implementations show that the protocol is loop free, does not block links, is fully transparent to hosts and neither needs a spanning tree protocol to prevent loops nor a link state protocol to obtain low latency paths. Implementations in Linux and OpenFlow on NetFPGA show inherent robustness and fast reconfiguration. Previous simulations showed a superior performance (throughput and delay) than the Spanning Tree Protocol and similar to shortest path routing, with lower complexity.

The demo is focused on the implementation of ARP-Path (a.k.a. FastPath) bridges, a recently proposed concept for low latency bridges. ARP-Path Bridges rely on the race between broadcast ARP Request packets, to discover the minimum latency path to the destination host. Several implementations (in Omnet++, Linux, OpenFlow, NetFPGA) have shown that ARP-Path exhibits loop-freedom, does not block links, is fully transparent to hosts and neither needs a spanning tree protocol to prevent loops nor a link state protocol to obtain low latency paths. This demo compares our hardware implementation on NetFPGA to bridges running STP, showing that ARP-Path finds lower latency paths than STP.