“There was no denying the crispness of the cable: flexing it exposed the copper of the conductor within”
Steve tackles spontaneous computer combustion and the incoming effect of 5G in business networks, plus a mix-up over moos and moons
For only the second time in my entire career, I’ve seen what looks remarkably like a case of spontaneous computer combustion. As with the previous incident, this was a desktop PC displaying a narrow swath of extreme heat damage starting from the central connectors on a standard SATA power plug.
Unlike my last experience, the owner of this machine was convinced his device wasn’t driving something with a heavy power draw. Its only external connection was to a CD drive, which looked clean and tidy to me. He’d followed the wisps of black smoke to the offending PC – that being the first sign of trouble. In common with most office PCs these days, the machine wasn’t busy at the time. It should have been in a sleep state, not maxing out the power supplied to a device that most of us haven’t had cause to use much this decade.
There was no denying the crispiness of the cable: flexing it exposed the shiny copper of the conductor within. While the interior of this PC was clean, the efficacy of the sleep-states and power management meant that the standard heatsink and fan were finely covered with an insulating blanket of fluff – a standard result of running these parts at very low temperatures and airflow rates.
The question this raised was, just how much heat energy is given off by a wire burning well enough to leave black soot trails on nearby metal surfaces? However much power that was, it wasn’t noticed by the machine’s thermostatic fan controllers.
Neither I nor the client had an appetite for endangering other bits of equipment by swapping around parts to see if anything else could be induced to catch fire. The PC still functioned at the motherboard level – so the power supply was working and nothing had overheated on the motherboard.
Rather irritatingly, this machine and every other item in the building with electricity passing through it was adorned with PAT testing stickers. This test might make the owner feel he was covered, but it wouldn’t extend right down inside the guts of the machine – and it certainly wouldn’t regard a 12-volt power feed as a likely source of an electrical fire.
To be fair, chances are that even if there was a test for such low-power cables, it wouldn’t provide a useful early warning mechanism for such an outcome. And for those who have ever seen and then acted on a SMART disk error warning – I don’t think it has any probes that look at power at all.
Is there a sensible response to this problem? I’m tempted to say no, because I haven’t seen a high enough incidence rate to make me feel anyone should be investing in some power super-controller idea, or whatever you might think would achieve genuine protection and no false positives. I also haven’t heard disk-recovery firms say that they only wish they could have got to the dead device before it set fire to its host PC, which gives an indication of the frequency of incident. Most of the
time, disk failures are inside the controller or in the spin or head motors. Power is the category that slips everyone’s mind.
The strange world of 5G
Making contact with big corporations can be frustrating. Just before the heavy snows of early spring, I got excited about the prospect of coffee with a CEO: an insider in the cellular business. This is a rarer type of contact for me than you may expect. No-one in the computing sector can understand why the telephone and cellular firms haven’t gobbled us all up in a takeover and merger orgy. One of the key reasons that this hasn’t happened is the infrequency of relevant contact. Telephone companies just don’t exist on the same planet as us.
And in case you’re thinking that this is just a serving of CSG (Cassidy sour grapes), I offer you the news snippet that Falcon Industries, Nokia and Vodafone have a joint venture to put a cellular phone network on the moon. I’ll pass over all the resulting jokes – I am, for instance, quite sure that this won’t support PAYG customers – and move on to the strange ways that this announcement came to light.
The proposed Lunar network, you see, is going to be 4G. In the weird world of public relations, this is being interpreted as a counter to all the noise being made at Mobile World Congress about 5G. This is where all the techies start sighing heavily, shaking their heads and frowning, because 5G most certainly isn’t the end of the road for 4G; it simply isn’t deployable in the same way, for one thing.
Why am I interested in this, if it’s so much at odds with the comparatively simple worlds of networks and business Wi-Fi? I started paying attention some years ago, when the Libor rate-fixing scandal sent waves through the finance community. Part of the reason for this situation reaching the scale it did, was because the conversations between the traders in different firms were undetectable to the companies in question. For one thing, they happened in public chatrooms; for another, they took place on smartphones talking over the public cell data networks, not via Wi-Fi provided by the building owner. Since that incident, the marketplace in connections has moved on – but only in a way that makes this problem more difficult, not simpler.
And, as this excitement about networks on the moon shows, there’s plenty of life in the old 4G dog yet. Especially in the area of MIMO. This is the tech that improves the throughput of a cell basestation by allowing multiple conversations between phones and the antenna. When I first heard about MIMO, I was shocked - like most networks people, I left single-conversation turnaround behind a long time ago with the demise of coaxial cables for Ethernet. It was a surprise to find that the cellular guys were only just getting into that area in the second decade of the 21st century.
Fast forward to 2018, and I was in a watercooler planning conversation with an IT manager, faced with setting out a medium-term infrastructure plan for a semi-rural site. Of course, he said, most of his planning was about wireless: how high he could make his mast; how fast the network would run. And whether bring your own device would even be possible, because it would be a continual fight between the compatibility of consumer devices such as tablets and phones, and the security features and threats presented to a raw Wi-Fi interface.
I encouraged him to stop researching that kind of thing, and to look instead at 5G. What if the convergent trends in networking were best addressed by admitting that designing everything to withstand routing over public services was the most likely architecture? Certainly, a whole lot of businesses like the idea of BYOD, and pretty much demand that employees use their personal connectivity to remain available 24 hours a day. So what’s the exact motivation for business-grade, old-school, large-screen deskbound computing or its infrastructure at all?
The answer seems to be in two wildly divergent directions. 5G is a short-haul technology. In pursuit of speed increases that were less mandated by the standards used, and more by the expectations of operators and governments, circling enormous bandwidth auctions back when that seemed like a good way to measure mobile service quality.
The jump from 4G to 5G was driven solely by people harping on about reported download speeds. By the time the whole specification/bidding stage had finished, the technology in people’s pockets had already leapt ahead, and made a lot of those early design decisions irrelevant.
But a standard is a standard, so what we’ll get from 5G is close-up, high-speed, low-penetration signals. Perfect for, say, a football stadium. Not so good for a tin shed full of men welding a giant metal structure, nor for a large estate of modern houses. It’s quite likely that almost any kind of office or industrial building will block a 5G signal from anywhere but next door. Existing neat workarounds such as cellular repeaters that plug into your ADSL router and take the traffic up via the internet can be expected to enjoy a renaissance – although it may then be that the ADSL used is nowhere near adequate to the demands of 5G phones and, of course, users.
The crossing point here is the thorny issue of traffic analysis. Nobody imagines that traffic analysis of public mobile phone network data is possible or desirable, yet the whole business of analysing traffic inside networks has taken off in the past 18 months – incredibly, taken off in the
“As the excitement about networks on the moon shows, there’s plenty of life in the old 4G dog yet”
sense that people are realising this could be a big, insuperable problem.
Say you’re a site manager, and you have a lot of 5G repeater boxes piggybacking on your internet link – it turns out that a member of staff (or even a van driver sitting in your car park) has been passing through your routers while engaged in a criminal activity. What will you say when the comparatively crude traffic management of phone packets delivers policemen to your front door?
The boundary here will become blurred. On the one hand, the short-range nature of 5G will increase the number of small repeaters put in by eager adopters. On the other hand, the headroom for more development in 4G, to handle less dense, more long-range orientated connections, will make the boundary between “inside, safe company network” and “outside, retail-grade, slightly less-safe cellular connections” much harder to identify and work with.
At the same time, BYOD machines will be far more expensive for the average worker to justify. I’ve heard one oft-told story about a hospital worker, who was horrified to find that only after she got the job did she get the instructions for connecting to the hospital intranet – which only worked on an iPhone costing £800.
Of course, regulations can catch up quickly with this problem. Everyone works with the antediluvian concepts of mileage allowances – personal vehicle versus company use – and yet there’s almost no regulatory management of the costs arising from a job that mandates internet access without costs to the employer. Nor yet has the regulatory dancing around 5G between cell firms and governments figured out the exact costs and benefits of the short-distance transmission problem to the ordinary consumer, employee, business owner or, indeed, network services designer.
While people bemoan the loss of network neutrality as a guiding principle in the USA, this is always going to have some positive outcomes in the field of working out how on earth your multiply-smart, multiplywireless, multiply-authenticated devices are actually connecting to the rest of the world. Being able to say “this network isn’t only used for business purposes but is also entirely transparent and aware of the data that passes through it” is a requirement that isn’t coming down from the ivory tower of a hopeful, blue-sky regulator. It’s the guys on the front line who are being given these questions to answer, alongside the techies from various sectors (antivirus, firewall, software-defined networking) being asked to meet the brief.
Telephone companies aren’t involved in this stuff. Their past approaches to the issue of “what passes through my network” have been about criminal activity, under-age access to adult material, and forensic traces of traffic that looks kosher but is implicated in a fraud.
None of these areas have given much comfort to those who think they provide a fix for something from the real world. The fact that some home internet access providers are specialised subsidiaries of old-school telephone businesses isn’t a clarification, either.
I’m pretty sure that the CIO of a company with large premises – an open-cast mine, for the sake of argument – will find it difficult to use 5G as a wider-area network to support their employees with both business and personal connections. Why? Because the person who signs off the firm’s statement on data security will take a dim view of services that aren’t transparent to later enquiry on what everyone has been chatting about.
This is the real problem standing in the way of telcos getting themselves into shape for the modern internet, or sensible staff members being able to show they’ve been clean with their personal devices exposed to the work environment. What happens when someone makes allegations about an infection or a data breach, and who should be justifying their actions to the ICO?
It’s an exciting time to be in this business, because the answers are clearly still developing. I just hope they develop quick enough to keep us all out of trouble.
Mooncall or Moocall?
In one of those classic teeth-grinding moments that the technology sector seems to specialise in, I very nearly wrote this month about the Moocall Calving System ( moocall.com).
It’s an IoT system that tells farmers when their cows are in the early stages of calving, by measuring the angle of dangle of their tail. A big green collar goes around the tail, right at the top; bearing in mind what else happens in that general area on most cows, I’m happy to say that I wasn’t given the chance to examine this tech close-up.
Inside the collar is a battery, some phone hardware and an inclinometer. While I’m far from an expert on the risks of cattle in labour, I assume there are benefits to a bit of early warning for everyone involved, starting with the cow, her offspring, the farmer and, I guess, the vet too.
It’s a great example of IoT and an area where cellular connections, battery life, infrastructure required and outcomes all make sense together. Cows on Wi-Fi is a bit of a challenge in terms of range, after all.
However, it wasn’t the best story of that wintry week. While I was busy reading the Moocall site and trying to select a cow picture, other parts of the PR machine were gearing up to talk about a cooperation between Elon Musk’s Falcon Industries and Vodafone, to land a cell network on the moon. As stories go, that one probably suits PC Pro readers better than something that could end up resembling an out-take from Withnail and I.
Why oh why did I end up with the agricultural example, when next door in the out-tray of the business was something about ROCKETS to the MOON? It couldn’t have been anything as simple as the difference between “mooncall” and “moocall”, could it?
“The shortrange nature of 5G will increase the number of small repeaters put in by eager adopters”