Seven factors that will hamper enterprise IoT adoption


The Internet of Things is today’s hot buzzword but out in the real world these seven factors will create roadblocks on the path to enterprise IoT adoption

Internet of Things concept drawing © bakhtiarzein - FotoliaBig Data and M2M (machine-to-machine) are yesterday’s buzzwords. Today all the hype has gravitated to the Internet of Things, which McKinsey says could account for an astonishing 11 percent of the global economy by 2025.

Whenever technology hype takes hold as virulently as this, it’s only a matter of time before overinflated expectations give way to sullen disillusion (a progess codified as the Gartner Hype Cycle).

So maybe this is a good time to consider the factors that will stoke disillusion by acting as a brake on enterprise IoT adoption. Based on careful reading of the research as well as listening to informed debate on the topic over recent weeks and months, here’s my list of the seven issues most likely to keep IoT at bay.

1. Too many platforms

Preparing to present a brief market overview last week at an IoT event, I assembled a non-exhaustive list of vendors who offer some kind of IoT environment. It’s a list that includes most of the leading technology vendors of our time:

This is a bewildering array of platforms all competing for dominance. They span several layers, from the Internet of sensors that collect data from physical devices and environments, up through the analytics platforms that assemble and analyze the resulting data, to the applications that take this output and produce consumable outcomes. Some are more focused on the consumer market, others are primarily concerned with industrial and B2B environments.

In the past, this degree of fragmentation into multiple frameworks and platforms would have caused huge problems of interoperability. As Internet of Things guru Kevin Ashton pointed out to me the other week, these days we have the computing power to mediate easily between multiple platforms and therefore there’s no longer so much of an interoperability challenge.

But there is a skills problem. It’s not realistic for people to become familiar with dozens of different frameworks, platforms and languages. So long as there are so many different alternatives to choose between, the wide range of choice will become a gating factor on any of them being broadly adopted. Many enterprises will hold back until some consolidation and convergence of these various platforms takes place.

2. Bandwidth constraints

One of the things I found interesting at the recent ServiceMax conference I attended in Paris was that many of the early adopters of IoT solutions in the B2B market are starting out in environments where a high-bandwidth Internet connection already exists. Commmenting on my report, Armelle Boisset, project leader at Kapsch CarrierCom, made the point that there are important constraints in most industrial environments:

There is a widening gap between existing networking-ready systems and older more standalone installed bases … Prior to analysing and acting on data streams, these need to be generated in the first place, with concurrent power and connectivity (not to mention security) constraints.

Lack of network bandwidth was a point reiterated by Dermot O’Connell, executive director and general manager of Dell EMEA OEM Solutions, speaking at a EuroCloud UK event on IoT which I chaired last week. He said that Dell has recently introduced a new product range, the Edge Gateway 500 Series, specifically to provide an interface from older machine-to-machine data sources into Internet of Things data collection. Data has been collected in industrial and commercial environments for several decades. But the devices that have been collecting that data were not designed to be Internet-ready. The Dell device is designed to bridge that gap.

O’Connell said that bandwidth constraints and security considerations mean that it is often impractical to transfer the entire data output from on-site devices into the cloud. The Dell device is designed to act both as a secure gateway and as a staging post where data can be filtered so that it only forwards the data that needs to be sent to the cloud.

Even though it’s becoming progressively cheaper to add sensors and intelligence to all manner of devices, the cost of connecting them up and transmitting the data will still make many potential IoT use cases too expensive to be economically viable. Budding IoT entrepreneurs should focus first on use cases in environments where bandwidth is plentiful, such as data centers and digitally connected workplaces and homes.

3. Insecure devices

While it’s true that there’s nothing new in the industrial landscape about adding intelligence to machines (cars have been built by robots for more than a generation now), they were never designed with connection to the Internet in mind. Like every other asset designed at a time when connections to the outside world were scarce and costly, they assume their local environment is safe from external threats. There are many newer devices too, especially in the home, whose designers never spared any thought or budget for Internet security.

Any Internet of Things strategy needs to take into account the potential risks of opening up such devices to the wider Internet — including inadvertent attachment when they are wirelessly advertising their presence to any nearby network.

The difficulty here is that the risk vector is largely unknown. Who has an asset register of every Internet-connectible device within their organization? Even if you had such a register, it would not protect you against undocumented vulnerabilities or unforeseen permutations of risk. This is the kind of uncertainty that is poisonous to innovation, and will lead many enterprises to simply shut down IoT projects rather than run the risk of some unexpected calamity. Products like the Dell gateway or GE’s Achilles industrial firewall play to these fears and may help allay them. Any IoT solution provider must have credible answers to these security worries.

4. Privacy concerns

Any time that connected devices interact with consumers or employees, the spectre of data protection rears its head — especially in European markets. Any data that contains ‘personally identifiable information’ (PII) is subject to data protection law. This not only means it must be carefully protected against unauthorized access, it also means the individual concerned has rights to know what data is being held and in some jurisdictions may have the right to revoke permission for it to be used. There may also be requirements to store PII in specific geographies.

All of this imposes extra costs for storing and managing the data, as well as introducing future uncertainty as data privacy laws evolve over time. Data protection compliance issues have been known to shut down enterprise projects at any stage, once lawyers are brought in to assess the risks. This looming shadow will persuade many enterprises to steer clear of IoT projects that monitor consumer or employee activity, even when the business case seems compelling. On the other hand, projects that concern themselves solely with collecting data that relates to machines or other non-human physical assets (anything from premises to cattle) can fast-forward without such worries.

5. Data hoarding

Another speaker at last week’s EuroCloud event argued passionately for enterprises to make their IoT data openly available in an eBay-like marketplace. Mark Wharton is technology lead at Iotic Labs, a startup that is building such a marketplace for IoT data to feed into an innovation space for IoT apps. Open sharing of data is fundamental to his company’s business model, so it’s hardly surprising he would argue for this. But he makes the valid argument that it’s only by connecting disparate data sources that innovative IoT applications will emerge:

The Internet of Things, if everyone puts their data into silos, nothing will happen.

Certainly we have seen useful new apps appear in the transport sector as a result of publicly available data that tracks aircraft, train or bus locations as they ply their routes. Wharton described a use case that had mashed up data from smart bee hives to environmental sensors and weather data to save water in a crop irrigation system.

Unfortunately, all the bandwidth, security and data privacy concerns we mentioned above, paired with an innate reluctance to share data, will lead enterprises to keep tight control of their data. IoT innovation will inevitably suffer as a result.

6. Inadequate data

In a conversation earlier this month with Twila Osborn, a vice president of R&D at Schneider Electric, I learned about the steps the energy management vendor is taking in product development to prepare for the Internet of Things. One point that stood out was the importance of making sure to collect the right data from devices.

You can’t just assume the product is going to deliver what you want from it. You have to take the marketing objectives you want to achieve in the product design.

If we don’t have the characteristics and the attributes we can’t build them in. If we don’t think about R&D, we won’t design in these attributes that are needed for these new business models.

The corollary is that, if the devices are not equipped to measure the data and attributes you need to produce your desired outcomes, then the data may be of little value. Osborn and her colleagues have found they need to think carefully about the data ontology at the design stage. Many enterprise IoT projects forged in the white heat of today’s hype may well founder once they discover the data they’re collecting can’t deliver the anticipated results.

7. Infrastructure in search of a problem

The final obstacle is the tendency of IT people to build out a technology infrastructure in response to the latest trend without first investigating what the business case will be. Those with long enough memories will remember some of the disastrously wasteful enterprise IT projects to build out service-oriented architectures, while in more recent memory there are the many initiatives to construct elaborate private cloud infrastructure.

Many organizations will now embark on projects to build out Internet of Things infrastructure that will end up languishing underused because no one thought through what its purpose would be. These white elephant projects will then contribute to disillusion with IoT even though they always were destined to fail from the start.

On this note, Dell’s O’Connell gave good advice on a three-step approach to embarking on IoT in the enterprise:

  1. Start with what you have — build on your existing technology investments and find some early, quick wins
  2. Architect for analytics — look for results from analysing the data you’ll be collecting to produce useful business outcomes, and be ready to scale rapidly into production if your pilot is successful
  3. Put security first — secure data at rest and in transit, and keep privacy and compliance front of mind

A fourth step, based on the advice from Schneider Electric’s Osborn, is planning to iterate through each of these steps as you refine your data models and analytics in the light of experience.

My take

Don’t say I didn’t warn you — but if you can surmount these obstacles, your success will be well deserved.

Image credits: Internet of Things concept drawing © bakhtiarzein.

Disclosure: Salesforce and ServiceMax are diginomica premier partners. ServiceMax paid my travel expenses to attend its MaxLive event in Paris and was a sponsor of last week’s EuroCloud event. I recently carried out a paid engagement for Dell. I act as volunteer chair of EuroCloud UK, to which my consulting business provides some event management services.