5G is a Delicate Balance
Part 2: What is more important, an intelligent core transport or a smart edge?In this second part of my latest series ‘5G is a delicate balance between’ I would like to take a deeper dive into the great debate: Smart edge versus intelligent core transport. The subject has been in such focus, that I actually had the honor of participating in exactly this debate at Mobile World Congress 2018.
See my MWC selfie here.
So what does 5G need to succeed?
More intelligent transport core or a smarter edge? While this has been a contentious debate, one thing is obvious, for the massive amount of data delivered on 5G, in line with so many service profiles, today’s networks will have to change. But will it mean bypassing the core and shifting all processing to a smart edge? Spoiler alert! No, for 5G to be successful, you’ll need both -- the right combination of smarter edges and a more intelligent core.
Before we get into the details, let’s explore theses concept a bit further. In today’s networks, most of the analysis, processing and decision-making is done in the core. For example a simple google query is said hit between 700-1000 machines, travel over 1500 km at the speed of light, to provide the user X million answers in a number of milliseconds. Load balancers, mixers and a variety of DNSs are all involved in this seemingly simple act of ‘internet search’.
Multi-access Edge Computing and the Smart Edge
However, 5G’s requirements will place great pressures on the network. To achieve the low-latency required for applications like autonomous vehicles, it will no longer be possible for data to travel to the network core for analysis and decision making. The distance traveled across the network, in itself, adds latency. And in the case of autonomous vehicles, anything but real-time is obsolete. The answer, at least in part, will require MEC solutions at the network edge. The concept, also known as multi-access edge computing, means shifting this computing of traffic and services away from a centralized “core” to the edge of the network and closer to the customer.
But is this enough to fulfill the 5G promise? I argue that this is not the case. Firstly, we can’t just shift everything away from the core. Many services, especially non-latency sensitive services, as well as data and information that needs to be exchanged between the edge nodes (e.g. to support mobility) will still be required to traverse the network to the core for processing. Some sort of central intelligence and management is needed, not only to oversee operations and direct the network elements, but also to load balance. Secondly, if all data would be processed at the edge, it will be extremely costly to implement MEC capabilities. Third, and not least, many services have an irritating habit of rising up on one side of the network and ending up on the other side. All of these patterns require connectivity over the core which is both dynamic and assured.
One might think that the current ‘data centric’ paradigm, using dynamic virtual overlay would work for connecting edges together, but overlaying a dynamic virtual network over antiquated static transport is bound to fail, as virtual overlays lack the assurance that is essential for premium 5G services.
So yes, we need smarter edges with more access points to handle the amount of analysis, data and processing for everything from high-bandwidth, low-latency applications all the way to voice and machine-to-machine communications. Therefore, finding the right balance of smart edge and intelligent core is required.
Making the Case for a Smarter Core
Today’s notion of just a single mobile core will be obsolete in the age of 5G. There will need to be multiple, distributed, intelligent NG core instances to optimize the processing and routing of information. Smarter core transport is imperative for end to end network slicing, a key component which will enable the wide range of use cases and profiles required under 5G. For example, connecting internet of things (IoT) devices, such as smart meters, on a network slice with highly availability and reliable data-only service at a specific latency, data rate and security level. Alternatively, for something like remote surgery, it can deliver another network slice with extremely high throughput with super-low latency.
As you can imagine, there are so many profiles and decisions that need to be made instantaneously, this will be impossible to monitor and adjust in near real time. Therefore, 5G will need a transport core that is enabled by artificial intelligence and machine learning. Yet, alone an intelligent core is not enough to meet the service level agreements (SLAs) and quality of service (QoS) needed for the long-term growth of 5G.
A More Intelligent Core and Intelligent Edge
In the end both an intelligent edge and a smarter transport core will be required to get the most out of our 5G network. 5G will require us to empower this vision of an intelligent core by providing some offload and relief with a much more intelligent edge. We have explored the topic of building artificially intelligent, multi-gear networks to deliver service assurance for 5G and beyond earlier, and this still rings true.
Advances in artificial intelligence and machine learning are making complex network management tasks achievable to empower this new realm of 5G including IoT, smart cities, and intelligent transportation systems. To do so will require us to have truly dynamic, optimized, and assured network.
This is a significant departure from the way networks are built today but can be implemented over time to maximize current and future resources. In closing, the future will no longer be the clear distinct separation between the edge and core we currently have, it will just be one equally intelligent network which, when required, moves workloads to the optimum position.