Future Networks, Services and Management

Enhancements in computing and growth in bandwidth, along with virtualization, have been driving force for the new applications and services in recent years. AI/ML technology is still in its infancy. We expect that to be one of the driving forces of future applications and services. As we move toward the year 2030, quantum computing should become a dominant factor for changes in devices, applications, networking, and services.

Towards the year of 2030 and beyond, many novel applications are expected to emerge as others mature, leading to increasingly intertwined human and machine communications. New applications often trigger new services and introduce challenging requirements that demand the continuous evolution of networking technologies. Thus, the inherent capabilities of interconnected networks and the running principles therein need to be enhanced, or even replaced, as requirements unfold. To help identify the essential network requirements and shape the future networks’ design paradigm, the authors have considered and reviewed a broad range of studies on the future applications and services from the recent years’ public research articles and reports. A number of typical applications are chosen to represent the trend. We also have a brief summary at the end of this chapter to provide an overview of application requirements.

Network services are the built-in capabilities that network infrastructure owners provide as an interface to end users and applications. Therefore, enabling the applications envisioned by Network 2030 will require to devise new services able to cope with the novel requirements. These services will not substitute the ones already in place, but rather they will complement them and improve the overall capabilities. This chapter deals with the definition of future services section. First, it covers limitations in the current network services and their inability to deploy emerging applications. Then, we provide a classification of the envisioned future services. The remainder of the chapter deals with the description of the different services. The chapter is finalized by a discussion of open questions and challenges.

The current Internet derives mainly from the 1980s and soon after. Among the key objectives were best effort connectivity and simplicity along with the ability to survive some level of link and node failures. Private networks have been used for applications requiring more assured security and privacy, and service quality better than best effort. Transmission rates were in kbps. Nowadays, the rates are in Mbps and Gbps ranges.

New wireline and wireless technologies are pushing the transmission rates from Mbps and Gbps to Tbps. Advances in space technologies are expected to make the space communications a viable alternative to wireline communications. In parallel to the advances in communications technologies, the number of connected devices and traffic is expected to grow to 100 billion devices and 175 ZB (i.e., 175 × 10²¹) by 2025, respectively. In addition, applications requiring large bandwidth and strict performance are growing rapidly.

As a result, the future networks will consist of many types of integrated networks and no longer be a vehicle only for best effort connectivity, but a programmable infrastructure of connectivity and applications supporting vital and high-precision services that require low latency, appropriate security, and extremely high reliability for communications between most of the locations in the world.

The intelligence is no longer only in the end devices but rather distributed among end devices, data centers, cloud, space, edge, and core devices in the network. As a result, the complexity is increased. To help deal with this increase in complexity, the automation of operational processes for inter- and intra-networking is being worked in the industry. On-demand modifications of network elements and applications are becoming a common trend. The level of intelligence in each component is increased with the proliferation of artificial intelligence (AI)/machine learning (ML) techniques, and advances in memory and computing technologies. By 2030, it is expected that self-managed networks will be available, with substantial user controls and tremendous growth in the services supported by autonomous edge devices.

This chapter divides networks and services into underlay and overlay networks and services. After describing underlay and overlay wireline and wireless technologies, the chapter provides insight into future advances in these areas by 2030 and beyond. Furthermore, the chapter proposes an architecture for future services, a management architecture, and describes application programming interfaces (APIs).

Since the dawn of the computer age, the amount of data our machines can process continues to grow relentlessly. To efficiently process applications such as AI, connected vehicles, AR/VR, and smart factories, computation must move ever closer to data that grows ever more massive. In computer science, this is a fundamental principle known as “data locality.” In future networks, we know it as edge computing.

In this chapter, we cover solutions to challenges posed by edge computing in future networks. We define the relationship between edge computing and 5G and MEC, and interfaces to Telco core networks and Hyperscaler public clouds. We look at access and edge components, including the “far edge.” From a business perspective, we identify revenue-driving applications and services, and latency-sensitive services vs. data sensitive. From a technical perspective, we describe edge architecture, including the concepts of an edge border gateway and federation (both edge-edge and edge-cloud), and key enabling technologies.

Modern large-scale data centers are very different than they were just a short time ago, which form the core infrastructure support for the ever-expanding cloud services. In this chapter, we survey representative data center network topologies, highlighting their advantages and disadvantages in terms of network architecture and scalability. Then, the data forwarding and routing schemes that are designed for these topologies are presented and compared based on various criteria. Thereafter, we discuss the traffic optimization techniques that are designed for the efficient operation of data center networks. A full understanding of the state-of-the-art data center networks is beneficial to future network design. We discuss several future trends of data center networks that let network infrastructure meet the ultimate challenges of the upcoming days.

Underneath the great innovation that is cloud, the glue that has made cloud technology possible is the network. Ethernet, Internet Protocol (IP), Transmission Control Protocol (TCP), and User Datagram Protocol (UDP), all other networking protocols that have been around for decades, are all still alive and well within the cloud. The only difference between building datacenters on-premises yesterday and building virtual datacenters in the cloud today, outside of the ease of use and programmable interfaces, is layers. Considering the Open Systems Interconnect (OSI) model, cloud networking is no different, just with more depth and additional layers to consider, not layers of the OSI model, but rather layers of infrastructure management as well as abstraction of the user space from the infrastructure itself.

Future Internet will utilise numerous methods of communication, including an increasing amount of space-based Internet transport infrastructure. Control and communication across Earth-based space-based networks present several problems—high dynamicity, spatial connectivity, continual movement tracking and prediction, ocular obstruction, integration with existing Internet infrastructure, routing, and addressing—all of which challenge existing control architectures and protocol mechanisms.

This chapter provides an overview of near-to-mid-term space networking towards 2030; it outlines the key components, challenges, and requirements for integrating future space-based network infrastructure with existing networks and mechanisms. We highlight the network control and transport interconnection and identify the resources and functions required for successful interconnection of space-based and Earth-based Internet infrastructure.

Finally, we discuss the management implications of these integrated assets and resources and potential technologies and capabilities that may be applied or extended.

Network slicing is a paradigm through which different virtual resource elements of common shared infrastructure (in both connectivity and compute substrates) become allocated to a specific customer who perceives the resulting slice as a fully dedicated, self-contained network for it. The resources are virtualized through a process of abstraction of lower-level elements, providing great flexibility and independence when allocating specific elements to the customer. This process permits the exercise of advanced actions such as scalability, reliability, protection, relocation, etc., along the network slice lifetime, without impacting the customer service. All these possible actions represent an incredible asset for a novel way of service provisioning with respect to the conventional mode of network operation.

Network slicing, despite not being a new concept, acts as a foundational concept and systems to current 5G/future networks and service delivery, with the goal of providing dedicated private networks tailored to the needs of different verticals based on the specific requirements of a diversity of new services such as high-definition (HD) video, virtual reality (VR), V2X applications, and high-precision services. Network slicing is supported by the technological paradigms of software defined networking (SDN) and network function virtualization (NFV) in an integrated manner. These three concepts, SDN, NFV, and slicing, encompass the overall trend of network softwarization, governing the transformation of operational networks. All of them will form the basement for the evolution of the network towards 2030.

This chapter starts with an introduction of classic routing protocols, followed by examining different addressing techniques. A key set of principles and requirements for routing and addressing for future Internet are discussed in detail from different perspectives, and special considerations are given to routing security and resilience. A few novel routing protocols, RIFT, LSVR, and SCION are introduced. Finally, there is a conclusion to summarize the discussions and areas for further research.

Network QoS refers to the mechanisms employed by routers and switches along the traffic path to manage throughput, loss, latency, reordering, and jitter of the traffic. Today, the Internet and many other TCP/IP networks only support the so-called best effort characteristic for traffic, which is insufficient to support the requirements of many current and, in the opinion of the authors, even more futuristic applications including real-time signalling and control, critical reliability, and application requiring any form of guarantees. While TCP/IP has seen a range of architectural options to support better than best effort service characteristic, these are often either limited in scalability, challenging to operationalize, or inflexible.

This chapter gives an overview of the current best practices of existing QoS mechanisms for TCP/IP networks, discusses gaps, and describes their applicability to different scopes of networks, such as the Internet, Home, Access-Provider, and Mobile Networks. It then suggests a longer-term evolution of the network scopes and discusses how to apply QoS in them. It then introduces a set of future QoS concepts including experience based and high-precision QoS. To enable such future QoS concepts, a future “toolkit” of architectural concepts is required in future networks, including programmability of QoS, virtualization of QoS, flexible network packet header functionality, instrumentation, and monetization.

For a 10 Mbps Ethernet link, a 1.5 Kb frames consumes 1.2 ms transmission time. This size is generally reasonable for traditional text or email type of applications. Today’s core networks are approaching 400 GbE bandwidth in the core networks and in order to fully utilize this kind of link rates, packet sizes need to be much higher. This also serves new types of applications well since the basic data processing unit of 4K/8K video frames, artificial intelligence (AI), and big data analysis is in the order of 10s of MBs.

This chapter introduces burst forwarding architecture. The burst paradigm deals with packing of a burst of several consecutive packets of an application data processing unit. It specifically addresses high bandwidth and low latency requirements of certain applications, e.g. holographic type of communication. This chapter firstly presents the use cases and identifies the problems of using packet switching. Secondly, the theoretical study of the burst forwarding technology is described in the aspect of network throughput, host data processing performance, data transmission latency, and router buffer requirement. Finally, the burst forwarding network architecture is described. The data packing, data forwarding mechanism, and flow control mechanism are described in detail. The burst forwarding technology is an end-to-end system which optimizes the network performance for the high bandwidth applications.

In order to guarantee secure operation of Network 2030s inter-domain infrastructure, we pursue a systematic design approach in this chapter. This systematic approach starts with concretizing the security properties that the network infrastructure should fulfill. In particular, we identify improved trust, path control, source authentication, and availability guarantees as the most important such properties. After laying out these goals, we articulate the constraints under which the properties must be achieved, e.g., the pervasiveness of (D)DoS attacks and the resulting need for lightweight security systems. Against the background of these goals and requirements, we propose a suite of security systems that comprehensively address current and future network-security challenges. More precisely, we present a path-aware network architecture that can be enhanced with a source-authentication mechanism, a bandwidth-reservation system, and a privacy-increasing forwarding scheme. Finally, we focus on practical considerations and expound the new services needed to implement our system proposals.

Network providers are under significant pressure to automate network operations to the greatest extent possible. Drivers for this include networking economics as well as the necessity to scale operations to keep up with the explosive growth of network size, complexity, and services. In the past, this has led to the pursuit of the vision of networks that are autonomic and self-managing and to technologies such as policy-based management based on the articulation of abstract operational rules independent of device details. The latest development in this space concerns intent-based network management, or simply intent-based networks (IBN). IBN is based on the recognition that even fully autonomic networks will still require guidance from operators to fulfill their intended purpose—from providing specific instances of services to achieving overarching operational goals under consideration of certain tradeoffs. IBN allows users to manage networks by defining expected management outcomes, as opposed to having to specify precise rules, steps, or algorithms that will lead to those outcomes. This requires systems that support advanced man-machine interfaces and that possess the necessary intelligence to identify the required steps on its own, generally using advanced algorithms, control loops, and artificial intelligence techniques. The following chapter provides an overview of IBN and its basic underlying concepts.

This chapter explores how different types of artificial intelligence can be used to enhance and improve network and service management. The addition of appropriate types of AI into a particular task (e.g., managing telemetry) is perfectly reasonable and can enhance both the performance and management of a task. It enables the organization to gain experience in incorporating AI. However, the true advantages of using AI will not be realized until AI is allowed to function as a system.

This chapter will progressively build an example of how AI can be incorporated into a cognitive architecture, currently being prototyped, to improve its network and service management capabilities.

Quantum technology is a rapidly advancing field which will revolutionize computing and communications networking. The use of quantum technology exploits the characteristics of quantum physics. In the quantum physics realm, subatomic particles don’t follow the same set of rules as the objects we can see and touch. These differences allow new paradigms to be developed for computing and communications. Quantum computers can be used to solve previously unsolvable problems on a much larger scale. The Quantum Internet can exchange large amounts of data using quantum physics properties thereby reducing traffic on traditional communication networks. This section will detail some of the unique aspects of quantum technology.