The digital transformation effects a wide range of industries including an increasing number of digital companies that do not deploy their own infrastructure
The rapidly evolving digital transformation has given rise to a new paradigm across a wide range of industries. This includes an ever-increasing number of highly successful digital companies that do not deploy their own infrastructure.
This revolution in Information and Communications Technology (ICT) is changing the value chain of all industries. Specifically, two fundamental aspects have changed:
- A dramatic digital business transformation enabling unprecedented efficiency through process automation.
- A radical shift to a services economy enabled through cloud-based delivery that result in significant pressure on traditional product value and cost.
The ability to support other industries is a major challenge for service providers going forward. Different needs apply to different industries as they go digital. Future competitiveness will rely on a much broader set of capabilities impacting company strategy. Entirely new business models will be possible through advances in connectivity, software, mobile devices and cloud.
The future network will need to be able to provide for different usage scenarios: high to low capacity, wide to local area, very dense coverage to spotty coverage, private to public, indoor and outdoor etc. We usually refer to these usage scenarios as the 5G use cases. This calls for a scalable, flexible and a modular approach when creating the network. The scalability will be able to go both upwards and downwards depending on the strategy chosen by the owner of the network.
It will be very important that the network provides interfaces that are built with business needs as the primary driver. Open interfaces that support a flexible approach to building new services will provide for a shorter time to customer.
The cost of acquiring and installing a telecom network must be taken into consideration, as well as the cost of operating the network. It calls for automation, machine learning and simplification as well as supporting the important use of abstraction to hide potential architecture complexity.
There are a couple of areas which are important to focus on to be competitive in the transformation to the future new ecosystems:
- Adaptive network quality must be enabled to handle very different use cases. The industry is moving from a consumer-led, smartphone-driven network era to increasing complexity, volume and diversity of requirements. Use cases range from extreme availability and performance (critical machine type communication) to scalability and volume (massive machine type communication).
- Efficient spectrum management will be more essential than today because spectrum will become an even more scarce resource. System coverage and capacity will continue to be a key driver of network design expanding to new higher frequency bands in licensed, licensed shared and unlicensed scenarios.
- Customer service optimization, both from cost and end user satisfaction perspective. Analytics and precise customer knowledge are essential to make this a mutually beneficial experience.
- Flexible business models. The final consideration is the ability for service provider to act in many different business roles, for example as a provider, supplier, customer, broker, etc. Each role dictates different requirements of functionality and business logic and each role requires different insights and segmentations for each business actor at the other end of the transaction.
With increased openness we also see an increasingly complex world with standardization organizations aiming to adopt to more flexible environments, for example 3GPP defining service-based architectures. We also see several open source initiatives in the communication areas such as ONAP and ORAN. All these initiatives aim for increased speed of evolution and reduced cost but also introduce multiple choices which can complicate network evolution.
To address these trends and challenges we have the following high-level view on the future network.
Starting with the transport network, software defined networking (SDN) is changing the way transport networks are handled, optical switching will over time take over more and more of the traditional networking and white boxes are slowly changing the routing and switching hardware. During the transformation, there will be a mixture of new and legacy equipment, so initially much of the new intelligence will be located to the management part of the network.
The cloud has come to transform how applications are built and managed. From the initial centralization of the cloud infrastructure there is now an increasing focus on solving the performance challenges through distribution of workloads across the network to where it makes sense to have them. The distributed cloud will span across the network and support many types of workloads. It is important to have a clear separation of concern regarding managing the basic infrastructure and managing the workloads running on top of it. It is however quite possible to create a homogeneous exposure of both workload API’s and cloud resources to other tenants.
When considering the access-mobility-network application layer, many of the earlier vertically integrated function are now being transferred to virtual machines and containers. New functions will in a flexible way be deployed where it is optimal from commercial, performance or other reasons and the future network will enable this. There will however be parts that for a foreseeable future will remain as native/vertically integrated deployments, e.g. the antenna near parts of the radio functions. The architecture supports network slicing which allows networks to be logically separated, with each slice providing customized connectivity, and all slices running on the same, shared infrastructure. Virtualization and SDN are the key technologies that make network slicing possible. Network slices are logically separated and isolated systems that can be designed with different architectures but can share functional components.
All these functionalities are managed by a highly automated management and orchestration layer. This layer handles life-cycle management, day 0 operations of the network slices and network functions, and day 1+ operation. The aim is to achieve close to zero-touch operations and to achieve this, technology such as machine learning, and artificial intelligence are used.
The end goal is to provide an automation stack where intents can be defined to describe the desired state of the network. This intent is translated by the automation platform in an optimized representation of the services and network resources which is then deployed on the network itself. From a networking perspective, this means to enable advanced closed loop automation logic to allow the automation platform to make sure the intent is always ensured. It is essential to provide a high degree of monitoring and control on the automation decisions. The future automation platform will need to provide a supervision enabled model of operation where the user can review and approve the proposed automation steps to build trust in the automation logic.
Intents will be expressed by a model and will be executed by the orchestration platform. Models are becoming the most important artefact describing automation and versioning of models will become essential. Strong Continuous Integration and Continuous Delivery (CI/CD) infrastructure is needed so that new versions of models can be verified and automatically deployed. It can be foreseen to have extended integrations between the vendors and customers CI/CD pipelines.
The management and orchestration layer will be where much of the network intelligence will initially evolve towards the zero-touch vision. It will also in an efficient and automatic way expose the network capabilities to other industries. The exposure and monetization will have to attract developers, tenants and the service providers and all will have to be able to innovate fast and make money. The architecture needs to provide simple and stable API’s that makes the whole network appear like a programmable entity. For the management & Orchestration part ONAP is now gaining momentum and will influence the evolution of this layer.
The future networks will utilize artificial intelligence to become a fully autonomous network with closed loop control and policy governance for dynamic behavior. The automation loops will exist on all levels of the network, from the extremely fast radio loops where the analytical data gets old in milliseconds to the cross-domain optimizations that predicts network traffic and load over long time periods.
In a similar way as for the network applications, the non-network applications benefit from the Distributed Cloud Infrastructure. There are several types of physical sites playing different roles, ranging from larger datacenters (national and regional datacenters) where the focus is on compute and storage, to medium sized sites (central office, local switching centers) where wide-area networking plays a greater role, to smaller sites (hub and antenna sites) which are optimized from an access networking perspective
Global connectivity and services have by tradition been deployed in a federated model, where the interfaces are well standardized and offered by one service providers. The complexity with multiple networks has been hidden through interoperability and inter service providers exchange models. However, the rapid deployment of new features makes the traditional standardized federated model hard to use. New methods of enabling exposure of assets from multiple networks is needed, like network asset facilitation and exchange or, on service providers request, aggregation into a single offer.