The rapid development and scaling of mobile telecommunications networks, together with related domains such as the edge-cloud continuum have raised significant concerns regarding energy consumption and environmental sustainability. Addressing these concerns requires a focus on CPU energy consumption, as CPUs are among the largest energy consumers in these systems. This paper investigates existing techniques, with a focus on CPU idle states (C-states), performance states (P-states), and frequency scaling governors implemented at both hardware and software levels. These mechanisms enable the dynamic adjustment of CPU parameters, providing opportunities to optimize power consumption, frequency, voltage, and overall system performance. In this regard, three CPUs with different architectures from well-known manufacturers, Intel® and AMD®, are thoroughly examined. A comprehensive dataset, collected under three load scenarios (idle, medium, and high), is used to support the analysis, reflect realistic runtime conditions, and enable a comparison of the technological differences in how these parameters are exposed and utilized.

The rapid advancements in technologies across the Computing Continuum have reinforced the need for the interplay of various network and compute orchestration mechanisms within distributed infrastructure architectures to support the hyper-distributed application (HDA) deployments. A unified approach to managing heterogeneous components is crucial for reconciling conflicting objectives and creating a synergetic framework. To undertake these challenges, we present NEPHELE, a platform that realizes a hierarchical multi-layered orchestration architecture that incorporates infrastructure and application orchestration workflows across diverse resource management layers. The proposed platform integrates well-defined components spanning network and multi-cluster compute domains to enable intent-driven, dynamic orchestration. At its core, the Synergetic Meta Orchestrator (SMO) integrates diverse application requirements, generating deployment plans by interfacing with underlying orchestrators over distributed compute and network infrastructure. In the current work, we present the NEPHELE architecture, enumerate its interaction workflows, and evaluate key components of the overall architecture based on the instantiation and usage of the NEPHELE platform. The platform is evaluated in a multi-domain infrastructure setup to assess the operational overhead of the introduced orchestration functionality, considering also the assessment of different topology configurations on resource instantiation times and allocation dynamics, and network latency. Finally, we demonstrate the platform’s effectiveness in orchestrating distributed application graphs under varying placement intents, performance constraints, and workload stress conditions. The evaluation results outline the effectiveness of NEPHELE in orchestrating various infrastructure layers and application lifecycle scenarios through a unified interface.

In order to satisfy the ethical requirements that are expected from upcoming 6G technologies, the knowledge of the power consumption ascribable to the applications and network functions is crucial to enforce a sense of responsibility and joint efforts towards value-driven sustainability. In this respect, this paper presents a modular, energy-focused prototype of the Management Data Analytics Function (MDAF), that represents the cornerstone of the Observability framework recently developed by the 6Green Project. Its main goal is to provide management data to upper layers (i.e., network and end-users/verticals). A 5G network was tested on both the User-Plane and Control-Plane; at the same time, said network was monitored by both the proposed MDAF and a common power monitoring solution: Scaphandre. Results show that the MDAF measures a higher power consumption than Scaphandre; the difference lies between 50% (when idle) and 4% (at the maximum offered load). This difference corresponds to the ”indirect” power consumption that the MDAF is able to ascribe to the containers/Virtual Machines (VMs) (i.e., the Network Functions (NFs)). The more accurate power consumption measurements of the single NFs is a first step towards the need to spread energy/carbon awareness to all the involved stakeholders.

The virtualization of UltraSound (US) medical imaging systems through cloud-edge computing offers a transformative approach to healthcare, overcoming limitations of traditional on-premises systems. By dematerializing US system functions and migrating them to the cloud, barriers related to hardware and localization are eliminated, enabling a more agile and scalable approach to medical imaging. This paradigm shift leverages virtual objects and composite virtual objects to intelligently manage and orchestrate physical and virtual components, optimizing system performance and meeting service requirements.

The emergence of 5G systems and beyond, along with state-of-the-art cloud and software engineering solutions, have brought to a radical shift of networking technologies with respect to the previous generations. However, the increased performance and flexibility have come at the cost of reduced sustainability, not only from an environmental standpoint but, in turn, affecting economic and societal sustainability as well. Infrastructure owners can introduce power management solutions in their datacenters, but the improvement is negligible unless tenant operating on top of the physical infrastructure are enabled and incentivized to customize their slices and their applications towards more energy efficient and carbon neutral profiles. For these reasons, the $\mathbf{6 G r e e n}$ project has designed an Observability framework for estimating the energy consumption and the carbon footprint of an entire network slice, or of edge-computing applications attached to it. This paper highlights the rationale behind the 6 Green approach and provides a high-level description of the observability framework.

A Packet delay emulator for high bandwidth traffic is a specialized tool designed to simulate network delays and measure the impact on high-speed data transmission. Simulating network delays and measuring the impact on data transmission allows network administrators and developers to test their systems in a controlled environment, mimicking real-world scenarios with different network conditions. This paper discusses the limitations of traditional packet latency emulators and the technical con-siderations required to test high-speed networks, and proposes a specialized packet latency emulator to replicate real-world network conditions. Results compare the proposed emulator with the best tools that are currently available and show a performance in line with the reference, with a significant accuracy improvement in the presence of higher latencies.

The availability of networks able to perform control and management operations at unprecedented speeds is a crucial achievement for supporting next-generation Industry 4.0 applications. In particular, the presence of humans interacting with moving objects calls for decisions to be performed at time scales such as to ensure achieving the specific requirements identified by standardization bodies for the industrial sector in Beyond-5G environments. Among these specific Key Performance Indicators (KPIs), network reliability must be carefully pondered to guarantee the safety of the human operators at any time while investing the proper level of resources. For this reason, this paper presents an optimization problem that finds the best set of redundant radio bearers for each User Equipment (UE) in the factory area while accounting for the resource usage. Such problem extends the current specification on redundant transmissions. Several heuristics have been designed to achieve the desired time scales that could not be fulfilled with an exhaustive search. Results show that the optimization algorithm and the heuristics provide the same outcomes in terms of selected bearers and loss rates, but the latter do so at significantly more stringent time scales.

Beside increasing flexibility and programmability, the current network “softwarization” trend is believed to be beneficial also in respect of energy efficiency, owing to the consolidation of resources made possible by virtualized networking components. However, the widespread use of general-purpose hardware may jeopardize energy saving, unless proper control strategies are put in operation. In this context, the paper addresses a “smart sleeping” control problem, where computing resources in multi-core processors executing network functions are modelled as multi-server queues, and the number of active processing units (either physical or virtual) can be dynamically adjusted by parametric control over a time scale compatible with the long-term dynamics of the traffic flows that require processing. We show that, on average, up to 25% of processing capacity of a network node can be turned off in the presence of bursty traffic with low load without significantly affecting packet latency.

In order to fulfill the stringent requirements of 5G applications, measuring the performance of the VNFs composing the network slices is crucial to identify potential bottlenecks of the networks. However, since the VNF behavior is time varying and strongly depends on the infrastructure and hosting execution environment, the traditional traffic generators are not suited for the evaluation as their overhead, both in terms of deployment time and code complexity, may affect the results to the point of corruption. In order to overcome this issue, this paper presents a software traffic generator, based on TRex and executed in a VNF, which leverages on an automation framework to provide zero-touch as-a-Service active monitoring. Results show that the impact of this solution on the measured performance is negligible in terms of deployment time as well as required input lines.

Service providers, 5G network operators and, more generally, vertical industries face today a dangerous shortage of highly skilled cybersecurity experts. Along with the escalation and growing sophistication of cyber-attacks, 5G networks require the training of skilled and highly competent cyber forces. To meet these requirements, the SPIDER cyber range focuses specifically on 5G, and is based on three pillars, (i) cyber security assessment, (ii) training cyber security teams to defend against complex cyber-attack scenarios, and (iii) evaluation of cyber risk. The SPIDER cyber range replicates a customized 5G network, enabling the execution of cyber-exercises that take advantage of hands-on interaction in real time, the sharing of information between participants, and the gathering of feedback from network equipment, as well as the development and adaptation of advanced operational procedures. This aims to help 5G security professionals improve their ability to collaboratively manage and predict security incidents, complex attacks, and propagated vulnerabilities. The SPIDER cyber range is validated in two relevant use case scenarios aimed at demonstrating, in a realistic, measurable, and replicable way the transformations SPIDER will bring to the cybersecurity industry.

Network softwarization has paved the way for 5G technologies, and a wide-range of (radically new) verticals. As the telecommunications infrastructure evolves into a sort of distributed datacenter, multiple tenants such as vertical industries and network service providers share its aggregate pool of resources (e.g., networking, computing, etc.) in a layered “as-a-Service” approach exposed as slice abstractions. The challenge remains in the coordination of various stakeholders’ assets in realizing end-to-end network slices and supporting the multi-site deployment and chaining of the micro-service components needed to implement cloud-native vertical applications (vApps). In this context, particular care must be taken to ensure that the required resources are identified, made available and managed in a way that satisfies the vApp requirements, allows for a fair share of resources and has a reasonable impact on the overall vApp deployment time. With these challenges in mind, this paper presents the Resource Selection Optimizer (RSO)– a software-service in the MATILDA Operations Support System (OSS), whose main goal is to select the most appropriate network and computing resources (according to some criterion) among a list of options provided by the Wide-area Infrastructure Manager (WIM). It consists of three submodules that respectively handle: (i) the aggregation of vApp components based on affinities, (ii) the forecasting of (micro-) datacenter resources utilization, (iii) and the multi-site placement of the (aggregated) vApp micro-service components. The RSO’s performance is mainly evaluated in terms of the execution times of its submodules while varying their respective input parameters, and additionally, three selection policies are also compared. Experimental results aim to highlight the RSO behavior in both execution times and deployment costs, as well as the RSO interactions with other OSS submodules and network platform components, not only for multi-site vApp deployment but also for other network/services management operations.

Multi-access Edge Computing (MEC) is expected to play an essential role in enabling 5G (and beyond) technologies and services. This has driven numerous micro-datacenter (μDC) deployment studies in the literature, with a common goal of addressing the optimal μDC placement and dimensioning problems. Along this line, this paper aims at clustering subareas with similar network activity dynamics, to find a good hotspots' representation over the urban area. Leveraging common Machine Learning (ML) and statistics principles, the main contribution of this paper is two-fold: (i) the definition and selection of dynamicity features based on real telecommunications datasets; and (ii) the granular cluster evaluation and analysis based on agglomerative hierarchical clustering. Three feature sets (containing 20, 12 and 8 features, respectively) are evaluated at varying precision levels, showing interesting trends on the number of clusters, heatmaps and intra-cluster correlation. These could potentially provide some valuable indications on the placement and dimensioning of the$\mu$DCs.

This paper aims to showcase the ability of the MATILDA Platform to enable vertical applications fully utilizing the capabilities offered by the fifth generation of mobile networks (5G). Although 5G, powered by network slicing and edge computing, promises to flexibly support radically new and extremely heterogeneous vertical applications, vertical stakeholders generally lack both the skills to exploit the full potentials of 5G networks and the vision of the underlying resources, owned by Telecom providers that are reluctant to expose them in an unmediated way. The MATILDA platform bridges the gap between the vertical application and the network service domains. This paper presents the case of an Industry 4.0 application, and highlights the role played by the MATILDA solution in its successful deployment and orchestration.

Since the adoption of virtualization paradigms is seen as a viable way to fulfil the requirements of next-generation applications in a sustainable way, this paper examines the virtualization of the Radio Access Network (RAN), focusing on the C-RAN architecture, in order to better understand the impact of NFV technologies on power consumption and costs in real networks. The evaluation compares the power consumption obtained by deploying the Base Band Unit (BBU) using commercial devices or pools of Virtual Network Functions (VNFs). Publicly available datasets describing the traffic and the eNodeBs have been used for the evaluation, as well as datasheets for both the commercial devices and the VNF pools. Results show that the usage of the virtualized BBU causes consumptions around 250% higher with respect to the commercial deployment, and operation and capital costs over 66% higher, contradicting the common belief of NFV being a “green” technology. Further estimates conducted in this paper, however, highlight how the deployment of VNFs alongside specialized hardware solutions can represent a successful approach for telecom providers, with energy savings up to 20% and costs in line with the ones of dedicated hardware deployments.

While the integration of fourth-generation (4G) networks with Edge Computing technologies would anticipate the improvements foreseen by the coming of 5G, as well as smoothen the transition to the new technology, 4G does not natively support Edge Computing. Therefore, specific functionalities for user-plane integration and isolation of tenant spaces are required for effectively deploying Edge Computing in 4G networks. This paper describes the design of the end-point between the mobile and edge environments that has been integrated in the telecom layer platform of the MATILDA Project. Such end-point, designed in a Virtual Network Function (VNF), allows intercepting and forwarding data and control traffic towards external Data Networks. Instances of this VNF can be horizontally scaled according to a decision policy, which determines the minimum number of instances required for the current load. Results show that the latency ascribable to the VNF processing is sufficiently low to satisfy the delay budget for all 5G use cases up to 10 ms and that the decision policy based on the QoS Class Identifiers (QCIs) allows scaling with the traffic load, while still fulfilling the performance requirements of each application.

A key trend of current network evolution is in the direction of network softwarization and virtualization. These technological paradigms aim to enable a network to be programmable in a way that makes the network more flexible, scalable, and reliable, and in turn leads to agile service deployment and lower capital and operational expenses. So far, two related widely adopted solutions are software defined networks (SDN) and network function virtualization (NFV). There is one main difference between these two new networking paradigms. SDN separates the control plane from the data plane through a well-defined programming interface, such that the centralized controller can have a complete view of the entire network, while NFV decouples network functions from dedicated physical equipment by means of virtualization technology, and runs the virtual network functions (VNFs) in the general purpose physical or virtual network appliances. Both approaches make the network programmable in order to have the aforementioned desired features. SDN and NFV do not depend on each other, and they actually complement each other. They can work well individually and can also work in tandem for performance reasons. Due to such advantages, both SDN and NFV have become key enabling technologies for 5G networks, and have also been used in a wide range of important areas including IoT, mobile edge computing, smart grid, cloud datacenters, and cognition-based networks.

Fifth generation (5G) mobile networks will lead to a deep integration between networks and applications. Through novel paradigms like network functions virtualization and edge computing, the new classes of heterogeneous application services will be enabled to run close to the mobile end-user devices with zero-perceived latency and fully-cognitive dynamic reconfiguration capabilities. Such “vertical” applications exhibit diverse performance/scalability requirements, and will rely on highly distributed, extremely virtualized, multi-tenant, and software-defined infrastructures. In such a context, handling the required operations in a scalable and dynamic fashion will be of paramount importance. A specific aspect, addressed by software-defined networking (SDN), regards the provision of suitable communication channels, once resource allocation mechanisms have performed the most efficient deployment of virtual network function instances, and VNF chaining needs to be implemented to enable network services. In this respect, this paper introduces the multi-cluster overlay (MCO) network paradigm: a tunnel-less SDN scheme for scalable realization of virtual tenant networks across the 5G distributed infrastructure, able to support (bulk) migrations of software instances among geo-distributed computing resources in a seamless and effective fashion. The numerical simulation and experimental results show that the MCO achieves up to over one order of magnitude smaller number of forwarding rules than the other state-of-the-art SDN mechanisms, while also assuring high performance during reconfiguration operations.

The upcoming Cloud-Fog interplay is expected to grant service providers more degrees of freedom in the implementation and management of their service portfolios. However, recent advances in Mobile Internet have developed a growing need to support user mobility - as users move, the Fog counterparts of services that require close proximity may call for migration(s) to meet the desired quality of service (QoS). With the state-of-the-art virtualization technologies, next-generation Cloud/Fog services are being implemented in modular software (i.e., as a graph/chain of portable virtual objects (VOs)) that can be migrated around the Telco infrastructure, and yet scalability is still an open issue, especially with the inter-datacenter bulk live migration of VOs. In this perspective, a VO clustering and migration policy that jointly considers user proximity and inter-VO affinity is proposed to scalably support user mobility, while allowing service differentiation among users. Results confirm that introducing migrations improve the QoS to always meet or exceed the requirements, as compared to static service placement, and considering VO clusters as aggregate entities will initiate around 40% less migrations, on average - an improvement that increases with inter-VO affinity and could potentially simplify service management when supporting user mobility.

To effectively support new cloud services and smart device accessibility, current Internet design must move away from its typical TCP/IP-based architecture and evolve towards new paradigms characterized by high degrees of freedom. In particular, Network Functions Virtualization and Software-Defined Networking have proven to be appropriate approaches to introducing edge-programmability. In this article we give an overview over the fog computing-architecture approach INPUT (In-Network Programmability for the next generation of Personal CloUd Service SupporT) with its implementation OpenVolcano, which uses the above approaches to promote future personal Internet cloud services at the network edge. Specifically, it enables computing and storage capabilities for edge network devices, provides a simple, private, programmable SDN-based Ethernet domain for end users and edge cloud services, IaaS and PaaS support with tight delay and performance constraints through templates. The focus of this work is on multi-service orchestration and monitoring, which is an essential core of any edge computing architecture.

Within a 5G mobile network scenario, this paper adopts Multi-access Edge Computing (MEC) and Network Function Virtualization (NFV) technological frameworks to evaluate the gain in terms of user Quality of Experience (QoE) that can be achieved when contents and server of a video delivery service move from the cloud at the edge of the Radio Access Network. The application scenario we set up envisions a video streaming service geared a small group of individuals who have strong social ties and who move in an urban space. The scaling up of this emergent type of social interaction is set to become in the near future highly demanding for the resources of mobile operators. The paper recreates the network infrastructure of a mobile operator in the city of Milano. We model network conditions through Mininet, and manage (computing, network and storage) resources by means of an orchestrator named OpenVolcano. The paper shows that, compared to the traditional cloud approach, a MEC one provides better QoE to group members simply taking advantage both of the proximity and distribution of mini data centers and of proper resource orchestration.

The Network Functions Virtualization (NFV) paradigm is undoubtedly a key technological advancement in the Information and Communication Technology (ICT) community, especially for the upcoming 5G network design. While most of its promise is quite straightforward, the implied reduction of the power consumption/carbon footprint is still debatable, and not in line with the energy efficiency perspective forecasted by the ETSI NFV working group (WG). In this paper, we provide an estimate of the possible future requirements of this upcoming technology when deployed according to the virtual Evolved Packet Core (vEPC) use case specified by the ETSI NFV WG. Our estimation is based on real performance levels, certified by independent third-party laboratories, and datasheet values provided by existing commercial products for both the legacy and NFV network architectures, under different deployment scenarios. Obtained results show that a massive deployment of the current NFV technologies in the EPC may lead to a minimum increase of 106 % in the carbon footprint/energy consumption with respect to the Business As Usual (BAU) network solutions. Moreover, these values tend to increase at a very high pace when the most suitable software/hardware combination is not applied, or when packet processing latency is taken into account.

In this demonstration we will show the prospects of dynamic cloud service placement. A cloud service is implemented that manages real-time video streaming and real-time control commands for a remote controlled drone. The requirements for this cloud service are groundbreaking because the image transfer and the control commands should be transmitted in real time that the drone can be controlled smoothly by a user. The goal is to improve user QoE for streaming services and real-time control by cloud service migrations, respecting the concept of dynamic Edge Computing by means of moving computing and monitoring applications on demand close to the user and the end device.

By bringing Cloud-like services much closer to endusers and their devices, Fog computing is foreseen to expand the scope of overlay networks, encompassing different heterogeneity dimensions (i.e., size, geographical distribution, devices/virtual objects (VOs) involved, quality of service (QoS) requirements, etc.). Although highly flexible solutions like Software-Defined Networking (SDN) have been conceived to handle such heterogeneity in future networks, scalability is still an open issue, especially with respect to Fog computing requirements. In this paper, we propose an SDN-based network slicing scheme for supporting multi-domain Fog/Cloud services, which offers high scalability, among other aspects, over legacy ones. Results show that the number of unicast forwarding rules needed to be installed in an overlay drops by up to over one order of magnitude and 4 times compared to the "fully-meshed" and OpenStack cases, respectively, at the cost of possible path sub-optimality, albeit knowledge on the datacenter topology can be used for VO placement optimization.

In the last few years, network “softwarization” is gaining increasing popularity to achieve dynamicity and flexibility. Cloud computing, as well as the new paradigms of software defined networking (SDN) and network functions virtualization (NFV), are supporting this evolution. However, the need to move services closer to users to guarantee low latency in the service fruition on one hand, and the trend to support personalization of services on the other, are stimulating the migration of services toward edge nodes (in the so-called “fog computing” fashion). This is the target of the INPUT platform, proposed in the INPUT project to support future Internet personal cloud services in a more scalable and sustainable way, and with innovative added-value capabilities. The INPUT platform enables next-generation cloud applications to go beyond classical service models, and even replaces physical smart devices, usually placed in users' homes (e.g., set-top-boxes, etc.), with virtual entities, providing them to users “as a Service.” In this paper, we present the INPUT paradigm and discuss a relevant use case-namely, the virtual set-top-box-adopted to prove the feasibility of the softwarized SDN/NFV paradigm jointly with the fog-computing approach for the support of personal cloud services. The INPUT platform is also compared with a legacy approach to evaluate the gain in terms of quality of experience for both static and mobile users.

The objective of this paper is to determine how the incoming and outgoing traffic patterns provided by the TCP impact the energy efficiency of a networked device under realistic scenarios. We set up a complex test-bed that allowed us to perform a large number of measurements of energy- and network-performance indexes on a general-purpose computing system with power management capabilities, and the Linux operating system. The performed measurements gave us a chance not only 1) to provide a complete energy profiling of TCP connections, but also 2) to analyze the role of underlying network protocols in detail, and 3) to identify specific situations where current network protocols may trigger high inefficiencies in networked hosts. Based on the obtained results, we also conducted further experiments for evaluating techniques for reducing the energy consumption induced by TCP data transfers under different scenarios.

The constant evolution and expansion of the Internet and Internet-related technologies has exposed the limitations of the current networking infrastructures, which are represented by the unsustainable power consumption and low level of scalability. In fact, these infrastructures are still based on the typical, ossified architecture of the TCP/IP paradigm. In order to cope with the Future Internet requirements, recent contributions envisage an evolution towards more programmable and efficient paradigms. In this respect, this paper describes the basic issues, the technical approaches, and the methodologies for the implementation of power management primitives in the context of the emerging Software Defined Networking. In detail, we propose to extend one of the most prominent solutions aimed at increasing networking flexibility, the OpenFlow Protocol, to integrate the energy-aware capabilities offered by the Green Abstraction Layer (GAL). However, the mere introduction of node-level solutions would be of little or no use in the absence of a network-wide management scheme to guarantee inter-operability and effectiveness of the proposed architecture. In this respect, this work also proposes an analytical model for the management of a network with these capabilities. The results will show how our solutions are well suited to provide a scalable and efficient network architecture able to manage the orchestration and consolidation of the available resources.

We propose an analytical model able to accurately estimate both power consumption and network performance indexes of Energy Efficient Ethernet (EEE) links working at the three available speeds under various traffic load patterns and packet size distributions. The model is sufficiently flexible and accurate to consider different traffic parameters; among others, the packet size distribution, the average burst inter-arrival rate, the burst size distribution, etc. With relatively low complexity, since it addresses stationary queue behavior, the analysis allows obtaining the average energy consumption of the link and, unlike previous works, the mean latency time experienced by incoming packets in closed form, without any upper or lower bound approximations. This aspect makes the model suitable to be adopted in optimization frameworks for network design and control purposes. The numerical results of the model are validated against measurements on a real test bench.

Multi-core processors optimized for networking applications typically combine general-purpose cores with offloading engines to relieve the processor cores of specialized packet processing tasks, such as parsing, classification, and security. Unfortunately, modern embedded operating systems still lack an effective and advanced hardware abstraction to exploit these aspects optimally. Based on these considerations, this paper proposes a novel framework, OpenFlow in the Small (OFiS), specifically designed to provide a flexible hardware abstraction layer for heterogeneous multi-core systems with advanced hardware accelerators for network offloading. OFiS represents such accelerators as standard OpenFlow switches inside the processor, moving the edge of the OpenFlow network management to the computational resources inside the end-boxes. As indicated in the experimental evaluation, OFiS exploits hardware parallelism and consolidates the software tasks at finer granularities.

With the aim of controlling power consumption in metro/transport and core networks, we consider energy-aware devices able to reduce their energy requirements by adapting their performance. In particular, we focus on state-of-the-art packet processing engines, which generally represent the most energy-consuming components of network devices, and which are often composed of a number of parallel pipelines to “divide and conquer” the incoming traffic load. Our goal is to control both the power configuration of pipelines and the way to distribute traffic flows among them. We propose an analytical model to accurately represent the impact of green network technologies (i.e., low power idle and adaptive rate) on network- and energy-aware performance indexes. The model has been validated with experimental results, performed by using energy-aware software routers loaded by real-world traffic traces. The achieved results demonstrate how the proposed model can effectively represent energy- and network-aware performance indexes. On this basis, we propose a constrained optimization policy, which seeks the best tradeoff between power consumption and packet latency times. The procedure aims at dynamically adapting the energy-aware device configuration to minimize energy consumption while coping with incoming traffic volumes and meeting network performance constraints. In order to deeply understand the impact of such policy, a number of tests have been performed by using experimental data from software router architectures and real-world traffic traces.

This paper deeply and experimentally analyzes the efficiency of low power idle techniques when applied to packet processing engines of network devices. To this purpose, we set up a complex testbed that allowed us to perform several measurements on energy- and network-performance indexes. The reference device platforms that we selected for this evaluation are new generation software routers based on component-off-the-shelf hardware, since they already include advanced power management capabilities, and can be considered as a significant example for next-generation green network devices. The results collected in the measurement campaign allowed us not only (i) to provide an in-depth energy consumption profiling of SR data-plane, but also (ii) to clearly outline energy costs due to the use of low power idle techniques. Among other interesting aspects, we completely characterized the energy consumption due to wakeup transitions, which may cause instantaneous consumption spikes higher than 4 times the power energy requirement when active.

Multi-core processors for networking applications typically combine general-purpose cores with off-loading engines, to relieve processor cores of specialized packet processing tasks, such as parsing, classification, and security. Unfortunately, modern embedded operating systems still lack of an effective support and hardware abstraction for optimally exploiting the above mentioned aspects. Starting from these considerations, this paper proposes a novel framework, “OpenFlow in the Small” (OFiS), specifically designed to provide a flexible hardware abstraction for a wide set of heterogeneous multi-core processors with advanced network off-loading capabilities. The OFiS framework allows SW services/applications, or only the operating system, activating and customizing off-loading operations, and distributing them to processor cores on a per-flow basis.

In the last few years, the ever increasing attention to the eco-sustainability of Internet has pointed out the need of providing advanced power management schemes in network devices, in order to be able to reduce energy consumption according to the traffic load. In this context, starting from the observation that most of the traffic of today's Internet is carried by TCP, we focused on the impact that the dynamics of the TCP congestion control may have on energy efficiency of end hosts. In this paper we demonstrate that great energy consumption is related to network congestion conditions. Moreover, we propose to use a technique, called Active Window Management (AWM), with the aim of driving TCP sending rate according to the network available bandwidth, so improving congestion control and providing high TCP performance. AWM requires no changes to any TCP version. We demonstrate that the TCP traffic shaping performed by AWM allows a great reduction of the energy consumption in end hosts, with respect to the case where only TCP is used as congestion control.

The energy management of computer hardware mainly relies on primitives defined in the ACPI industrial standard. Network devices can borrow the same concepts as a basis for their energy-efficient design; however, to effectively tailor network processors for their specific applications, the set of energy-aware states defined by the ACPI and their parameters should be carefully chosen. In particular when designing network hardware, one should address the problem of optimizing parameter values (most often chosen in a discrete set) to trade-off energy efficiency and network performance. In this paper, we provide a methodology for such choice, by investigating this trade-off over a continuous spectrum of parameter values. This approach will allow us to gain insight in the behavior of the optimal solutions that achieve different desired tradeoffs.

In this paper, we consider energy-aware network devices (e.g. routers, switches, etc.) able to trade their energy consumption for packet forwarding performance by means of DVFS techniques. We focus on state-of-the-art packet processing engines, which generally represent the most energy-starving components of network devices, and which are often composed of a number of parallel pipelines to “divide and conquer” the incoming traffic load. Our goal is to control both the power configuration of pipelines, and the way to distribute traffic flows among them, in order to optimize the trade-off between energy consumption and network performance indexes. With this aim, we propose and analyze a constrained optimization policy, which tries to find the best trade-off between power consumption and packet latency times. In order to deeply understand the impact of such policy, a number of tests have been performed by using real-world traffic traces.

Advanced power management capabilities have been proposed to be included into next-generation green network devices in order to modulate their energy requirements according to the workload. The clear side effect of enabling these new capabilities consists in a performance level reduction of network devices. Starting from some existing benchmarking standards for evaluating energy-efficiency, namely ECR and ATIS-060015, this contribution is devoted to determining a set of parameters, and methodologies that can be applied to correctly and precisely evaluate the tradeoff between energy consumption and network performance. Some experimental results obtained with the proposed indexes and methodologies, as well as a green SW router prototype have been provided.

In this contribution, we explore and evaluate the feasibility and the impact of power management policies, able to well suit a heterogeneous set of highly modular architectures, generally used for developing today's network equipment. The proposed policies aim at optimizing the power consumption of each device component with respect to its expected network performance. In order to provide an experimental evaluation of the proposed ideas, we applied such power management policies to a new generation SW router based on multi-core COTS platform, and we evaluated it with real traffic traces.

In this work, our main objective is to explore how Software Routers (SRs) can deploy advanced and flexible paradigms for supporting novel control plane functionalities and applications. To this purpose, we investigate and study a new open-source SW framework, namely Distributed SW ROuter Project (DROP), which aims at developing and enabling a novel distributed paradigm for IP-router control and management. DROP is partially based on the main guidelines of the IETF ForCES standard, and it allows building logical network nodes through the aggregation of multiple SRs, which can be devoted to packet forwarding or control operations. In addition to the original ForCES design, DROP will aim at extending router distribution and aggregation concepts by moving them at a network-wide scale in order to enable and to support value-added services for next-generation networks.

We consider a new generation of COTS software routers (SRs), able to effectively exploit multi-Core/CPU HW platforms. Our main objective is to evaluate and to model the impact of power saving mechanisms, generally included in today's COTS processors, on the SR networking performance and behavior. To this purpose, we separately characterized the roles of both HW and SW layers through a large set of internal and external experimental measurements, obtained with a heterogeneous set of HW platforms and SR setups. Starting from this detailed measure analysis, we propose a simple model, able to represent the SR performance with a high accuracy level in terms of packet throughput and related power consumption. The proposed model can be effectively applied inside "green optimization" mechanisms in order to minimize power consumption, while maintaining a certain SR performance target.