Were small devices and sensors ever connected before the term the ‘Internet of Things’ (IoT) was coined?
Sure they were, especially in industrial applications, where the term SCADA (supervisory control and data acquisition) sums up what all of this is really about—information flow and control. This comes from the 1960s, when computing was much more standalone—i.e. precious little in the way of interconnected networks and transports were closed and proprietary. While networks evolved and grew, the key for the revolutionary changes in connectivity was open systems, and a set of internet protocols (IP).
These tore through larger proprietary computing systems, headed down to the desktop and into the smaller devices mobile sector where network operators saw an opportunity and the term ‘Machine to Machine’ (M2M) gathered some momentum for a while.
The low cost of connectivity componentry and pervasive open, IP networking allows pretty much anything to be connected to anything. In the 1990s, the head of a large analyst firm said that by 2004 even light bulbs would have IP addresses.
Herein lies a potential problem—how do you give everything a unique address?
The IP standard for addresses most widely in current use—IPv4—has been creaking and groaning for some time as its rather limited addressing system has been used up. In 2011, the last blocks of IPv4 addresses were allocated to the regional Internet registries. By now, several regions have run out and grey markets exist for addresses.
The use of Network Address Translation (NAT) is widespread, rife and vital. Inside your own ‘private’ network, you run your own IP addressing system, and a server maps this via a single address to the internet as a whole. My light bulb can have the same IP address as yours—in fact millions of light bulbs can all have the same address without impacting the overall IPv4 address resource pool one iota. Without this, little would function.
This wasn’t quite how it was meant to happen. The Internet standards group, the Internet Engineering Task Force (IETF), created a new standard in the 1990s, IPv6 (don’t ask why IPv5 never made it to market). Apart from the astronomical number of addresses (3.4 x 10 to the power of 38—more than enough for not only light bulbs, but also tulip bulbs and many other things besides), it has capabilities that allow for seamless mobility, plug and play management, and better support for multimedia applications.
It would seem a ‘no brainer’, and despite a World IPv6 launch in June 2012 involving carriers, ISPs and networking equipment manufacturers, still IPv6 usage is operating at relatively low percentages of internet traffic. On December 9th 2013, just 2.24% of users were accessing Google over IPv6.
Will the interest around the IoT stimulate more action around IPv6?
It might help, but there are more pressing demands on the IoT than addresses and actually ‘things’ might not need to be on the internet after all.
In amongst all of the hype about how it is now easy and cheap to connect anything to anything, an important detail is often missed—why bother? Many business cases for the IoT are based on what is technically possible, and tend to gloss over who benefits and how much effort they might have to put in and over what timescale to achieve the benefits.
Just as an example, the oft-quoted poster child of the IoT is smart metering, yet even this is proving difficult to justify without legislative incentives. This should not really be a surprise as most consumers struggle to see the benefit for themselves, and as electricity meters generally have an installed lifetime of 25 years and cost little to manage, there is little there to attract mass updates from the utility companies. Mobile networks have hardly been around that long, so marrying the two into a new value proposition that is worthwhile for device manufacturer, installer, network operator and energy supplier is not easy.
Even when a connected device application proves worthwhile, there are potentially many more downsides if the connection is universal. There might be too much data traffic over the means of communication chosen, pushing up costs and potentially affecting the quality of service—and for SCADA applications (for example, within the nuclear energy sector), this is often far more critical than the speed of users swapping emails or quality of digital phone calls.
There are also risks around security and resilience, which might be better minimised by isolation and containment within an internal internet, where IPv4 is entirely sufficient, rather than opening up to a wider network with unknown consequences.
In truth, those beating the IPv6 drum cannot rely on the IoT to make their lives easier. The internet has a huge role to play in SCADA, but as a form of networking standardisation and not necessarily as a form of universal connectivity. Standards-based devices might be more easily clustered together in a way to build IP connected applications, but as an analogy to social media, in groups where they have ‘shared interests’ and are accessed or controlled by individual users who also share that interest—not an Internet of Things, but ‘Social Networks of Whatever’.
This article first appeared on http://www.computerweekly.com