By 2020, the analysts tell us that there will be 50 billion IoT devices, 212 billion sensors, and 44 zettabytes of data produced per annum. Depending on your point of view, the good/bad news is that, currently, 85 percent of the so-called smart sensors, are not so smart – they’re not connected! So, the ICT infrastructure needed to enable this digital world - where Big Data and the Cloud is joined by Artificial Intelligence (AI), Machine and Deep Learning(ML/DL) and Virtual Reality (VR) – is going to have to be drastically different from the present one. And if the ICT infrastructure is going to change, then it follows that the buildings which house much of the hardware and software – the data centres – are also going to have to undergo something of a transformation.
Few, if any, industries will be immune from this ICT revolution. If you work in the manufacturing sector, for example, you might think that you’ve already seen high levels of automation and robotics bringing high levels of efficiency and innovation to the factory. And, in plenty of cases, there’s already some kind of data analysis being implemented to help improve various stages of your business. At the simplest level, a customer order received at a retail outlet will be either fulfilled there and then, and a stock replacement list will be sent back to the manufacturer at the end of a day or week’s business; or the order will be sent through to the factory for fulfilment – with order tracking keeping all those involved (customer, retailer, distributor, manufacturer, shipper etc.) informed as to the progress and eventual delivery of the customer’s goods. Clearly, some kind of ICT will also already be an integral part of the product manufacturing process – from design (CAD) and production (CAM/CNC etc.) through to the logistics of shipping.
So, the manufacturing industry uses plenty of compute, networking and storage, housed in computer rooms and larger, dedicated data centres. Why the need for change?
Well, let’s flesh out today’s largely theoretical potential of IoT, AI, ML, DL and VR, and examine what possible impact these technologies might have on any particular manufacturing organisation.
You can’t sell a product you haven’t yet designed (a least, not easily!), and you can’t sell the designed product unless you have some customers. So, design, sales and marketing are all linked, and all can make massive use of, for example, virtual reality. Product design using VR takes the idea of CAD to a whole new level. And, once the product has been designed, well, what better way to sell it than by letting the customer experience it, again using VR? Your compute/networking/storage requirement just got a whole deal bigger – and that’s not allowing for the pop-up websites and social media marketing campaigns you want to fire up and run for a few days or weeks, before retiring them. Or texting an existing customer as they walk past your shop to alert them to a new product, or a special offer. Or requesting (and processing) customer feedback. A further requirement for more IT, and this time requiring a flexible, expanding and contracting on-demand resource.
Move on to the actual manufacturing process, and we’ll take for granted the robots, the just in time, lean manufacturing philosophy. What about the possibility of real-time monitoring and management of the manufacturing process? And the potential of accurate, predictive maintenance, so you replace plant machinery parts only when necessary, and not just in case? And longer term analysis of the overall manufacturing process, to see why/where problems occur and what can be done to eliminate these? Lots of sensors all over the factory, providing raw data that, properly processed, can add a whole new level of manufacturing optimisation and efficiency. More compute/networking/storage required.
Logistics operations already make use of technologies such as GPS, fuel efficiency software and the like, but virtually all of these report in to a human being. What if artificial intelligence and machine learning took over and were capable of providing a new level of logistics optimisation, spotting trends and efficiencies beyond the view of a human being? Yet more IT!
And we haven’t even reached the really clever bit yet – what happens once the product is delivered to the customer. It’s here where IoT’s potential is causing such excitement. At the most sophisticated level, an autonomous vehicle will deploy hundreds, if not thousands, of sensors which will a) ensure it operates correctly and b) provide ongoing, real-time feedback to the manufacturer. And even if you think that the internet fridge is pointless (and lazy!), it’s not difficult to think of just how many products, machines, buildings, humans will benefit from clever use of data generated by IoT sensors post-sales. Safety, product improvements, efficiencies, just in time maintenance…the list of possible benefits to be obtained from sensor-generated data is almost endless. All culminating in the relatively new idea of the Digital Twin – whereby a virtual copy of a physical product exists in the Cloud – a rather more sophisticated database than a simple spreadsheet! Your compute/networking/storage requirement is starting to get out of control, but the benefits of Complete Product Lifecycle management are hard to ignore – faster time to market, an increase in product quality, constant customer interaction and much, much more.
To achieve these benefits, not only will you need more IT resource. This resource will also need to be flexible, cost-effective and a mixture of centralised and distributed, or edge computing. Some data will reside and be processed in the main data centre facility, more and more data will be created, processed and even deleted locally. So, time to re-think your data centre strategy. Can you afford to set up the required, hybrid ICT infrastructure yourself – build/own/operate a central facility and lots of satellite, micro data centres? Or is the digital world’s required substantial increase in ICT best met by use of colos and Public Cloud providers? Or, a hybrid combination of your own and third party data centre resources? Perhaps the perfect ICT infrastructure is a hybrid, hybrid approach – centralised and local, in-house and outsourced?!