Some quick (perhaps not fully-formed) thoughts on the supply and management of private 5G networks for industry. These come from a series of conversations for an upcoming report (and webinar) about the kind of service level agreements (SLAs) that might go to under-write industrial-grade 5G. The report is being written because, well, it is just about all anyone is talking about, as old familiars in the telco market look to earn their sea legs in the blustery swell of Industry 4.0.
The question about industrial-grade 5G SLAs is nigh-on impossible to answer, it seems – certainly when asked about the futuristic industrial landscape that is supposed to be propped up by ultra-reliable low-latency communications (URLLC) sometime after 2023(?). As it stands, for all the fanfare, very little in the way of factory operations (if we take the core Industrie 4.0 concept, geared around smart manufacturing, somewhere in Germany’s industrial heartlands) is running on 5G.
But for some minor proofs-of-concept in ringfenced production lines, away from the real business of actually making things, there just ain’t that much happening. This is partly because there are hardly any industrial 5G devices available. Siemens has just released a 5G router. Lufthansa Technik, one of the few companies really rolling with this stuff, told an event a few weeks back that it had only just taken receipt of its first proper 5G unit, to attach to its twin private 5G setups in Hamburg.
As written, then, there is a whole discussion about who blinks first over the supply and demand of industrial 5G chips, between the likes of Qualcomm and Bosch, say – and that, in the end, neither will budge until the delayed standards-releases for 5G-proper come into view. So nothing much is doing. Despite all the press releases about private 5G deployments, there is a lot of thumb-twiddling in the Industry 4.0 ranks. Just about everything is on LTE. Ask Athonet or Quortus or Druid Software; this is where business is being done right now.
Ask Nokia, as well; visitors to its factory in Oulu, in Northern Finland, say the word on the shop floor is LTE is animating its production lines. 5G? Even Nokia’s factory engineers shrug a bit, the story goes. And Nokia’s marketing team, in the latest podcast in the Enterprise IoT Insights series on private wireless, says the same: 5G will come, and will matter, but 80 percent (or something) of industrial use cases can be served with LTE.
Which returns us to this question about SLAs for industrial-grade cellular: what is there to write down in a contract, really, when the technology is being used to either replace misfiring Wi-Fi networks or to play in the industrial IoT sandbox – or, tentatively, to run non-critical manufacturing processes? Because no one is betting their factory on cellular, yet. No manufacturing company is putting SLAs against 5G performance, yet – until the systems are tested and proven, and humming like a turbojet (to borrow a line).
And nothing is humming because nothing is available, in any volume, and in any serious industrial garb. Because 5G has not yet been fully developed and standardised. More than this, the question about SLAs is difficult – if not impossible – because the kind of use cases URLLC-grade 5G is supposed to enable are hard to discern, as yet. The job to construct a matrix in three dimensions to tally unknowable use cases against unknowable networking KPIs, against an unknowable Industry 4.0 market (even manufacturing is hard to pin down), is a futile one, at this stage.
What about SLAs for private LTE, then? If LTE is being deployed to upgrade parochial Wi-Fi systems, which top-out for latency and throughput, or fall-down as soon as a modicum of mobility is required, then it probably over-delivers. Certainly, Lufthansa Technik’s experiments suggest even LTE will do the trick, in some cases. The sense is LTE is being introduced in industrial settings as a curtain-raiser, ahead of the real spectacle; it is increasingly easy to deploy and manage, and the applications it is being applied for are the kind of everyday disciplines that should, just, fare better.
LTE is not burdened with the same weight of expectation. It works like a balm for today’s factory headaches and not in the anticipated style of 5G as a kind of industrialised version of Weird Science – where a dream factory is magicked-up on a computer. The point is, maybe just, that SLAs for edge-based LTE networks are not that hard to write, or might not be that much in demand. It works well, and is appreciated where it is set to work. It is a better than what has gone before, anyway. So what’s to worry about?
The other question in all of this – as 5G technology develops and 5G use cases emerge, and as network providers figure out what to deliver in terms of performance (and what to guarantee) – is about who manages the network itself. Of course it is; what else are we talking about, here? This is, arguably, the defining narrative for the troubled carrier community in the 5G era – alongside parallel tragi-comedies about commoditized hardware, open source software, and the lengthening shadow of hyper-scalers.
There are plenty on the quiet fringes of the telco sphere, right in the middle of the private networking bubble, that reckon mobile operators are done for, effectively. They have two things going for them: spectrum and networking know-how. The liberalisation of spectrum for ‘vertical’ pursuits in leading industrial markets – started in the US with the long process of CBRS re-farming; driven-through in Germany, in a heartbeat, with the 3.7-3.8 GHz carve-up; copied one way or another in the UK and Japan, notably – has robbed them of their monopoly on airtime.
Meanwhile, the move to shrink and simplify mobile networks in software has, arguably, dumbed-down the whole networking discipline. Suddenly, anyone can license airtime and manage a network. Suddenly, anyone can be a mobile operator. And anyway; who says Deutsche Telekom or Vodafone, or any other of the old dogs of telecoms can manage such a diverse customer base anyway? Running one national network, one way, for millions of customers, is very different from running a million networks, millions of different ways, for millions of customers.
Is it not easier for each of those customers to run its own network, or networks, for its own idiosyncratic ends? Well, yes and no. It’s too simplistic – and neat, in this Industry 4.0 bluster – to say that; such a deathly outlook is a symptom of an early-adopter market, where the freakish tier-one (tier-zero?) manufacturing brands are going about like a band of war-painted Schwarzeneggers with digital grenades to light a fire under the whole Industry 4.0 picture, while a whole army of extras plays in the smoke.
But as the smoke clears, we will see the cast of characters is only multiplying; the old faces are not exiting, or being replaced, neccessarily. There is a whole SELL / BUILD / RUN process that needs to be worked through, for every-sized enterprise in every-sided industry, which will find roles for every kind of supplier, including traditional mobile operators. It is perhaps hard to believe what the do-it-yourself low-power end of the market says about IoT as a team sport. It sounds naive. In the end, however, everyone, new and old, will play a role – it is just that some will just play more of a role.
But, what role for the old protagonists in the telecoms sector? Ah, but I’m out of time, with a newsletter to file; we’ll pick up again another day, if I ever get around to it…
This article is continued here, under the header, ‘Vendors sell, integrators build, operators run – the case for carrier-led private 5G (ramble on)’. Tune in to the upcoming webinar on Industrial-grade 5G SLAs on March 24, featuring ABI Research, EXFO, Metaswitch, VoltDB, and Zeetta Networks. Registration is open; sign up here.
