CSP Evolution

On course for a telecoms singularity

SDN '16 Superpanel - Duration: 2 minutes

The ‘Technological Singularity’ is the idea that when we distill ‘artificial superintelligence’ our technologies will be released for endless growth and improvement. Once our programs are coding ever more intelligent programs to improve upon and replace themselves, we humans can (and indeed must) stand back and enjoy the fruits of an artificial intelligence explosion as the improvement cycles speed up.

It’s a compelling vision but that technological singularity is unlikely to arrive any time soon. Meanwhile there is a less dramatic singularity brewing in telecoms which looks set to provide a big assist for telecoms industry transformation.

For the last few years the industry has been fixating on three developments: Network Functions Virtualization and Software Defined Networking (NFV/SDN), Internet of Things (IoT), and the development of a set of fifth generation mobile network technologies (5G). Taken together they promise to drive through a new chapter in the evolution of the public network and the Communication Service Provider.

We’ve come a long way and have achieved even more than we hoped

In this edition of CSP Evolution we survey the joins and dependencies of the three strands of communications and IT development – the ways they might ultimately prove to be highly dependent on each other for success and, just as important, where extra attention might be needed at the joins to ensure that they do.

For instance, without NFV as an underpinning, many believe that the more advanced use-cases and applications for 5G may be unable to be economically deployed and managed. Similarly, without fifth generation wireless technologies, our ability to serve what many expect will be an explosion in IoT use cases would be limited. Not only are vast numbers of ‘things’ expected to be connected, but they are certain to be highly differentiated in terms of their data demands – some ‘things’ will require only episodic polling at low speed, others will be required to burst sudden volumes of data (security cameras, for instance).

csp2_tracker_1_-_abi

Some applications will be happy to operate in batch mode, sending big chunks of data to a central repository on a regular basis, others will – in the future – require ultra-low latency to support high levels of interactivity. Such a list of requirements suggests a need for a diverse set of wireless network ‘types’ to serve them all with flexibility and agility.

We are definitely going to see, a major shift in CSP capabilities

Likewise the underlying network ‘platform’ (beyond the various radio technologies) will be based on NFV/SDN and must feature a very high level of agility and extreme automation. There is no way that IoT services – eventually involving billions of individually connected things – can be dependent on men with screwdrivers.

Which brings us back to the notion of a telecoms singularity. With all three of our telecoms technology strands working together and amplifying each other’s capabilities, we may be able to engineer a self-perpetuating "explosion” of our own.

NFV ISG: a long way in a short time with further still to go - 40 seconds

IoT and NFV

CHAPTER 1

Network Functions Virtualization (NFV) sees the telecoms industry adopting the virtualization techniques developed successfully for the data centre, but attempting to do it in a ‘telecoms’ way: that means a heightened focus on interoperability and standards, along with an insistence on high resilience and availability.

The idea is that telcos will end up with a highly scalable infrastructure based on commodity hardware (rather than expensive specialist ‘black boxes’) with control plane and data plane separated for maximum agility and the whole lot software-driven and highly automated. Of course there are a few problems. By its very nature such an infrastructure is designed to be generic, almost service-neutral so that it may run the diversity of known (and as yet unknown) services over it.

Is IOT an enabler for NFV?

But there is a persistent problem. What seems to work on paper and PowerPoint with NFV/SDN, (and also in the lab) doesn’t necessarily translate quickly into sales. Why? The fact is that it’s hard to work up a short term financial justification for what is – despite all the claims of OpEx and CapEx savings – a major investment.

csp2_tracker_2_-_ihs_technology

A business ‘transformation’ will always be much broader than simply buying in some capital equipment and new software – in fact infrastructure investment may be a small part of the entire effort with things like business redesign, process automation, staff training and so on being just as, or more, important than updated technology.

So any CFO worth his salt is entitled to look with great scepticism at an infrastructure transformation and ask hard questions about return on capital and even on whether the technical requirements of the network might be better served with incremental investments rather than an initially expensive step into the unknown with NFV and its attendant downside risks – basically the lingering fear is: what if it doesn’t work? NFV is a strategic play that will pay off over time, not a cost-saving solution to a specific set of problems.

Another reason for the agreed slowness of the uptake is what Marcus Brunner, Head of Standardization at Swisscom, calls the perennial "elephant in the room” at all the meetings. The pachyderm in question, he says, is the conflict (ongoing) between standards and open source. "The entire industry doesn’t know how to deal with this,” he says. But he’s also noted a change from a couple of years back when nobody knew much about NFV, to now: "It’s now very good and people work well together; there’s a trust between different stakeholders and everyone understands his own position and is able to compromise.”

So given all this, how does IoT, as one of our three pillars, help justify investment in NFV?

 

There are definitely use cases for NFV and IoT both outside and inside of 5G

According to Marcus Brunner, IoT is a lot about services so the end product of IoT is really the ‘digitalisation’ of the industry as a whole (not just the telco) and there will be a big role, eventually, for NFV in IoT. The real question is: who provides what bits?

Swisscom sees NFV as a prerequisite for the 5G journey - 50 seconds
ETSI NFV reaches out to 5G to get detailed specifications - 19 seconds

5G and NFV

CHAPTER 2

Fifth generation mobile (5G) is the next set of radio network technologies currently being researched, standardised and readied for production by around 2020, if not before. As is usual, some telcos seem determined to get to market pre-standard so some pre-standard 5G services are bound to be launched early to get a competitive jump.

Unlike its predecessor ‘Gs’, 5G is really a bundle of different radio network technologies – a so-called ‘heterogeneous’ network – with ultra-fast and low-latency radio technologies at one end of the spectrum and slow speed "narrowband IoT” (NB-IoT) at the other. Diversity rather than narrow standard setting is the name of the game here.

5G already has several species of IoT radio technology, for instance, and some telcos (such as Swisscom, Orange and South Korea’s SK Telecom) are using the non 5G standard LoRaWAN as well. As with earlier ‘Gs’, 5G will be as much about blending existing technologies (in this case LTE and even 3G) into the mix.

The diversity of use cases will be extensive and much of it will have to be part-enabled ‘behind’ the radio access network (RAN) where transmission services must be engineered to match use cases. Some services, it is thought, will require extra-stable low-latency conditions. Others will be optimised for video, others for ultra narrowband IoT services, yet others will be required to pitstop at the edge of the network so that mountains of data can be sifted and analysed rather than passed on to the core network.

The consistent message I hear is that NFV, SDN and MEC (mobile edge computing) will be part of 5G

One of the likely 5G requirements is expected to be services to support autonomous vehicles. These services are going to have to be both ultra-reliable and ultra-fast. So fast in fact, that they must do better than a millisecond on round trip delay, which means in turn, no long trips to the data centre. This implies that tendrils of cloud must be extended towards the network edge from where demanding applications can store data and respond quickly – perhaps to changes in the physical environment (a car pulling out from an intersection, say). That circumstance can be transmitted to all the vehicles driving autonomously on a particular stretch of road. It’s a capability called ‘edge’ or ‘fog’ computing and is expected to play a significant role in delivering next generation applications.

csp2_tracker_4_-_abi_research

Often mention is ‘network slicing’, a technique available through NFV which enables the 5G operator to create appropriate virtual networks for different services with different sets of quality guarantees. But some 5G operators, such as Vodafone, are also looking forward to using ‘standard’ NFV to better manage their networks: "To move network nodes around when we want.”

In talking to participants at the recent ETSI NFV #15 meeting in Sophia Antipolis, France, a picture emerges of an interest group fully aware of where some of the most compelling use cases for NFV will emerge. Of our three, three-pillar relationships, the one between 5G and NFV is the most compelling and the NFV Industry Standards Group is keen to start defining use cases with their equivalent numbers in the 5G ecosystem.

We got very interesting feedback from the business and technical perspectives

According to HPE’s Marie-Paule Odini, the September 2016 ETSI NFV meeting involved a dedicated 5G meeting with the aim of reaching out to 5G bodies and academics from all over the world. The objective, she says, was to get feedback from both business and technical perspectives and to understand the 5G use cases and requirements. "This is an ongoing dialogue with the long term overall objective being to get to interoperability with very detailed specifications,” she told us.

Susana Sabater, Manager, Cloud Network Designs at the Vodafone Group, is adamant that NFV and 5G are complementary initiatives. Susana says Vodafone is no virtualisation newbie – it’s been many years working with virtualisation and is currently part way through a huge network transformation based on NFV and work from the TM Forum, IETF (Internet Engineering Task Force) and ONF (Open Networking Foundation).

Vodafone’s experience so far with the ETSI version of NFV has led Susana to conclude: "We could deploy 5G without NFV but if we want all the benefits of 5G we need to have the virtualisation.”

ETSI NFV reaches out to 5G to get detailed specifications - 80 seconds

5G and IoT

CHAPTER 3

SDN '16 Superpanel - 80 seconds

5G and IoT are clearly going to be hugely important to each other. How does 5G influence the development of IoT, TelecomTV asked Matt Hatton of IoT specialist consultancy, Machina Research. "Well it’s top priority for just about every carrier and vendor in the world at the moment,” he says.

That’s because of the CSPs’ apparent existential crisis. If the network business is about selling users bandwidth, then any number of telecom consultants are able to map out an alarming reality about the industry’s long-term health. There will inevitably be a graph with a costs line going more or less sideways and a ‘revenue per megabit’ line heading down. The crossover point is armageddon. That’s when there’s not enough revenue coming in to fund the network and make a return for the business behind it.

In Orange’s 2020 strategy we identified IoT as a key topic/market

View full video: Eric De-Beau,Head of R&D, Network Division, IT Solutions for the Core Network, Orange

Clearly network competition is more nuanced than this and ‘things’ can be made to happen to get the lines moving in the right direction again. The most popular ‘thing’ is to start selling services which do more than just count bandwidth in and out and send out a bill. Enter 5G wireless applied to emerging IoT services.

csp2_tracker_5_-_abi

How is this going to work? Well on the face of it IoT doesn’t look particularly bountiful right at the moment. According to Marcus Brunner Head of Standardization at Swisscom, "with IoT the discussion is a lot more around the networking part,” he says. There’s a fight between LoRa deployment and NB-IoT which shows that something is stirring. But, says Marcus, his customers aren’t looking at IoT platforms just yet. The first thing they want to do is figure out what the business model is and what their applications look like. The platform and the network services they’ll need once that is decided is some way off yet.

The enabler for these ‘low hanging fruit’ IoT services is actually the very low cost required for both the connectivity and the tiny embedded sensors that are going to be implanted in all manner of devices – from the unashamedly consumer (dolls, toys, novelty items, kitchen gadgets) to the highly useful (traffic monitoring, health tracking and so on). What they all have in common is a need to become viable applications through low (and even lowering) cost of deployment. For that reason what we can see of today’s low-powered IoT services doesn’t persuade us that they’re going to be huge cash generators for the connectivity providers, even at large scale.

IoT has the potential and excitement to be that new service opportunity

The revenue has to be generated by making the connections valuable beyond the revenue to be derived by selling the data the applications need. In other words, the little chunks of connectivity required are like table stakes for the service provider – it’s about making money from the services, not from 50,000 connections multiplied by ‘$0.x per month’, but from aggregated services which provide the real value for the customer by doing clever things and unearthing valuable insights.

If that’s the case, when should CSPs move? According Ray Mota, CEO & Principal Analyst at ACG Research, for CSPs "wait and see is not an acceptable option,” he says. They have to start acting now. "We think that IoT has the potential to be the big new service offering for CSPs”, having conducted some large scale research into the problems facing them.

Ray says that the concern he has is that connectivity should not be viewed as the CSP’s only option.

Indeed, CSPs need to grasp the opportunity presented to them by the three strands of the Telecoms singularity.

Fixed network virtualization will enable them to build an agile core and access network infrastructure to handle the plethora of IoT use cases that will develop in the coming years.

Both low-speed LPWAN and high speed, low latency 5G services (and many others in between) will enable CSPs to offer IoT radio access (and fixed access) services alongside their consumer video, Internet access and business services.

And their customer-facing and IT skills will enable CSPs to knit their 5G and NFV capabilities together to ride the IoT wave and win a large and growing share of the projected US$1.3 trillion IoT market.

5G and IoT are clearly going to be hugely important to each other. How does 5G influence the development of IoT, TelecomTV asked Matt Hatton of IoT specialist consultancy, Machina Research. "Well it’s top priority for just about every carrier and vendor in the world at the moment,” he says.

That’s because of the CSPs’ apparent existential crisis. If the network business is about selling users bandwidth, then any number of telecom consultants are able to map out an alarming reality about the industry’s long-term health. There will inevitably be a graph with a costs line going more or less sideways and a ‘revenue per megabit’ line heading down. The crossover point is armageddon. That’s when there’s not enough revenue coming in to fund the network and make a return for the business behind it.

In Orange’s 2020 strategy we identified IoT as a key topic/market

View full video: Eric De-Beau,Head of R&D, Network Division, IT Solutions for the Core Network, Orange

Clearly network competition is more nuanced than this and ‘things’ can be made to happen to get the lines moving in the right direction again. The most popular ‘thing’ is to start selling services which do more than just count bandwidth in and out and send out a bill. Enter 5G wireless applied to emerging IoT services.

csp2_tracker_5_-_abi

How is this going to work? Well on the face of it IoT doesn’t look particularly bountiful right at the moment. According to Marcus Brunner Head of Standardization at Swisscom, "with IoT the discussion is a lot more around the networking part,” he says. There’s a fight between LoRa deployment and NB-IoT which shows that something is stirring. But, says Marcus, his customers aren’t looking at IoT platforms just yet. The first thing they want to do is figure out what the business model is and what their applications look like. The platform and the network services they’ll need once that is decided is some way off yet.

The enabler for these ‘low hanging fruit’ IoT services is actually the very low cost required for both the connectivity and the tiny embedded sensors that are going to be implanted in all manner of devices – from the unashamedly consumer (dolls, toys, novelty items, kitchen gadgets) to the highly useful (traffic monitoring, health tracking and so on). What they all have in common is a need to become viable applications through low (and even lowering) cost of deployment. For that reason what we can see of today’s low-powered IoT services doesn’t persuade us that they’re going to be huge cash generators for the connectivity providers, even at large scale.

IoT has the potential and excitement to be that new service opportunity

The revenue has to be generated by making the connections valuable beyond the revenue to be derived by selling the data the applications need. In other words, the little chunks of connectivity required are like table stakes for the service provider – it’s about making money from the services, not from 50,000 connections multiplied by ‘$0.x per month’, but from aggregated services which provide the real value for the customer by doing clever things and unearthing valuable insights.

If that’s the case, when should CSPs move? According Ray Mota, CEO & Principal Analyst at ACG Research, for CSPs "wait and see is not an acceptable option,” he says. They have to start acting now. "We think that IoT has the potential to be the big new service offering for CSPs”, having conducted some large scale research into the problems facing them.

Ray says that the concern he has is that connectivity should not be viewed as the CSP’s only option.

Indeed, CSPs need to grasp the opportunity presented to them by the three strands of the Telecoms singularity.

Fixed network virtualization will enable them to build an agile core and access network infrastructure to handle the plethora of IoT use cases that will develop in the coming years.

Both low-speed LPWAN and high speed, low latency 5G services (and many others in between) will enable CSPs to offer IoT radio access (and fixed access) services alongside their consumer video, Internet access and business services.

And their customer-facing and IT skills will enable CSPs to knit their 5G and NFV capabilities together to ride the IoT wave and win a large and growing share of the projected US$1.3 trillion IoT market.