5G: The rollout necessities
5G is the new generation of mobile networks. Each of the previous generations enabled faster broadband speed and played a role in accelerating mass adoption of mobile internet. The physical world is now connected more than ever.
These days, there is an immense hype around 5G that promises significantly faster speeds, ultra-low latency and capability to handle massive-scale connections. However, the essence of 5G is not just an enhanced version of the previous “G” but an enabler of new industries and verticals. It brings new use cases in smart buildings, automotive, retail, healthcare, virtual/augmented reality to name a few. And it will help realize a new world for consumers and operators. 5G is expected to converge the physical world with the Internet world and this will enable a significant amount of new business opportunities.
The connectivity type and demands differ widely between these use cases. There are diverse requirements for latency, throughput, availability, power, and volume of connections for each tenant. This brings a deep shift in how Communication Service Providers (CSPs) will architect, supply and manage their networks. The network demands on connected utilities, where some of them are buried in building basements for years and where low power is needed for efficiency, are different from those on connected vehicles. Concepts like Network Slicing and Edge Computing & Storage would be needed to tailor the demand for each tenant. And technologies like LPWA, NFV, and SDN will be essential to implement this shift.
Those technologies, alone, are not sufficient for rolling out 5G. The massively distributed, dynamic, and programmable infrastructure makes effective network management a complex exercise beyond the possibility of manual intervention. An autonomous lifecycle management fabric is necessary. Building, testing and deploying virtualized network slices also needs to be done in an efficient way.
Autonomous Life Cycle Management
NFV and SDN technologies, for instance, bring many challenges to existing network management realms. They add complexity and scale, different layers - SaaS, PaaS, IaaS - on top of one another, service turn over time, self-mobility in failure scenarios and multi-tenancy. Most of the existing network management realms are not ready to handle these challenges. They are siloed and inflexible. There is coupling of data aggregation, storage, and analysis in different pipelines which impedes introduction of new analysis capabilities. On top of that fulfillment and network control is not tightly integrated with the analysis part.
A new approach is needed to build an autonomous life cycle management fabric that decouples the collection of data from analysis and consumption. The autonomous life cycle management should be able to consolidate data from all available data stores in the CSP landscape across data center, network, applications and also support systems such as CRM. The fabric should also be able to process real-time events and situations and classify these events in order to build the problem’s root and common causes as well as the service impact of these events on the end customers.
As shown in the figure, the autonomous fabric should also implement closed loops of control between analytics and fulfillment policy enforcement. This automated orchestration and instantly-provisioned assurance leverages full automation of the life cycle management.
Managing Network Slices
A network slice is a piece of the network - often virtual and software-defined - that extends from the core down to the connected last point fulfilling a certain demand and quality of service. Building a network slice includes integrating a chain of virtualized core functions, data center (cloud and edge) resources, wireless resources, and programing the connectivity that connects the chain together. An extensive testing exercise should take place particularly for latency-sensitive use cases such as a network slice providing connectivity between cars and infrastructures.