Edge computing continues to dominate agendas at data centre conferences as analysts, operators and suppliers get to grips with the implications for the sector.
Edge computing will not materialise independently: its development is driven by the next iteration of mobile telephony, 5G. So Edge computing and 5G are mutually dependent and will co-evolve: 5G cannot deliver on its promises without ultra-low latency and high levels of interconnection, or in other words, without a large fleet of smaller data centres, physically located within the populations they serve.
Together they will deliver faster data processing (some observers predict that by 2025 over 20% of data will be processed in real-time), local caching and sharing of compute power, energy efficiency at both network and device level, resilience and security and optimal work allocation.
It’s clear that we are in the presence of disruptive technology that will usher in a new breed of infrastructure, change supplier-customer relationships and introduce different business models. How worried should we be? Will Edge take us back to a distributed computing model and make data centres redundant? The answer is no, or at least, not yet.
The history of ICT has seen the primary business model move from consolidated to distributed and back again: We started with mainframes and bureau services where computing resources were too expensive for all but research organisations and the largest corporations to own.
ICT became more commoditised with the introduction of the IBM360 in the 1970s, but the equipment remained monolithic: a computer wasn’t a device, it was a room. That all changed with the introduction of the PC in the 1980s and computing exploded into every business, home and school.
As digital technology became more pervasive and our reliance on it increased, it made sense to outsource the heavy lifting to specialists, so over the last decade or so things have consolidated again, with organisations moving IT function out of notoriously inefficient office cupboards and server rooms and into purpose-built facilities, often operated by specialist third-party providers and consumers relying on remote applications and storage.
Now Edge looks set to take computing right back out to the customer – the ultimate in distributed IT. The difference now is that, rather than the sector oscillating between distributed and centralised business models, the market demands both – and everything in between.
Each new iteration of computing, each new business model, has served (or perhaps more accurately, generated) a new type of customer. The rapid rise of consumer-facing hyperscale cloud companies has not been at the expense of colocation providers; in fact, their demand for space was responsible for much of the recent growth in the UK’s colocation market.
By further extending the existing customer base of business, government and consumer to include a whole new category: things, Edge should massively expand the future market for data centre services.
So, the key question is “How do we ensure that the sector is positioned correctly to take advantage of this transition? Like every other innovation from the electric telegraph to the Talkies, there will be winners and losers. Which operating models are likely to be most easily adapted to support these new requirements? Are the risks and opportunities shared equally across the sector? What should providers be doing to prepare? The answer, predictably, is “it depends”.
We know that to realise the full potential of 5G the prerequisites are a lot of investment in network (upgraded connectivity with more cells, higher density and plenty of fibre), a lot of investment in compute (lots of smaller scale data centre capacity close to customers) and more spectrum.
Within the data centre, network upgrades will be needed with changes to switching and routers. Interconnection will be key, and we will hear a lot of talk about Network Function Virtualisation (NFV) and Software Defined Networking (SDN).
This all suggests that most of the short-term growth opportunities are within and around the telecoms supply chain and small-scale data centre construction and repurposing. However, as 5G densities increase, opportunities for existing data centres should emerge.
Analysts at Credit Suisse1 and ING2 predict that retail data centres, especially those with high levels of interconnection, will be best placed to benefit, but that Edge presents limited opportunities for traditional wholesale colocation providers unless they are sites that aggregate core network capacity.
On the other hand, Edge is also unlikely to disrupt or displace their existing customer base significantly. Our transition to 5G, and the development of the supporting Edge data centre infrastructure will take several years and there are barriers in the way.
Firstly, the mobile operators have to identify a business case (which can be tricky bearing in mind that the UK market does not generate the kind of returns that many other countries do) that would justify offering 5G outside urban centres.
There are also technical issues: UK masts are shorter than the European average which reduces range significantly (although the Government has just proposed easing restrictions on taller masts). Planning restrictions and the multiple and often conflicting requirements of many local authorities can dramatically slow down the pace of rollout: England alone has 155 Highways Authorities. And then there is the customer.
The benefits may seem clear to us: Operational efficiency will take another huge stride forward as will capacity: connection and download speeds will rocket, continuing the trend to transmit larger and larger packets of data, mostly video.
So far so good for the consumer, who will be able to download films in seconds, and for the worker, who can become genuinely mobile, taking their office with them via portable devices and collaboration tools. But the real value adds are likely to be outside the consumer market.
5G is optimised for small packets (the infrastructure is efficient enough to make huge numbers of small data movements economically viable) so has the potential to be a game-changer for other sectors, such as transport, connecting autonomous vehicles with roadside infrastructure, making smart cities real, enabling the internet of things.
However, this needs a high degree of customer imagination to perceive the business opportunities and/or cost savings and develop new business models to monetise these innovations. “5G is the accelerant that our nascent Edge computing environments demand. We are at the rough equivalent of 1992 with the internet.
By 1995 enough barriers to entry were lowered that when combined with a larger audience the internet had gone from novelty to changing how the world works.
The Edge without 5G is similar to 1992 internet and within 3-5 years when 5G has become more pervasive, we are likely to see new use cases that exceed the early growth of the internet.
In fact, I expect Edge to exceed the internet in scale, impact on businesses and our daily lives within the next 10 years” – Mark Thiele, Executive Director, Edge Cloud at Edge Gravity by Ericsson.
Vijay Mistry, from VIPA Digital summarises, “In a nutshell, to realise the potential of 5G three things need to be in place: robust interconnection capabilities, compute capacity adjacent to customers, and spectrum. Edge and 5G are mutually dependent and will co-evolve.
“There are also regulatory, technical and investment obstacles to overcome before 5G can reach its potential and maximise opportunities for Edge computing. The rollout sets a timetable for supply chain deployment, with short term opportunities in infrastructure, followed by network equipment upgrades and software.
“Most direct growth opportunities are within the telecoms supply chain but retail colocation providers with large interconnect businesses are best placed to benefit. The wholesale colocation customer base is not expected to change significantly. However, the transition will take several years, so there is time for operators to prepare and adapt”.
Source – Data Economy