Unlike most ICT models, the Internet of Things (IoT) arguably started as a ‘bottom-up’ concept: its inventors foresaw massed ranks of connected objects and devices before thinking through the connectivity and middleware that would manage comms links, devices, data, and application development. Filling the gap, IoT platforms emerged as the hierarchical layer that directs the ‘things’ as they fulfil their roles as efficiency improvers and revenue enablers.
IoT platforms can be configured to facilitate interaction between endpoint ‘things’, and to provision more complex infrastructures for distributed computing and the development of distributed applications. Described as a ‘central backbone’ for the IoT, they are assuming an important part in the establishment and deployment of IoT systems and applications.
As such, IoT platforms have become a market sector that’s hot, and going to get hotter. A forecast from IoT Analytics sees the IoT platforms market growing at a CAGR of 33 per cent up to 2021, with annual revenue topping $1.6 billion over the same period. The market analyst notes that while the markets for many IoT hardware components (such as sensors) and IoT connectivity solutions (such as 3G) have been around for years, the IoT platforms market was basically nonexistent before 2013. An estimated 300 vendors describing themselves as IoT platform providers are addressing this nascent product category.
“Enterprise platforms or software backend solutions are technically able to manage IoT data,” says Pádraig Scully, VP of market research at IoT Analytics. “However, these platforms were not designed with the sole purpose of running on network protocols such as MQ Telemetry Transport or Constrained Application Protocol, and OSs such as Raspbian or Brillo [as IoT platforms are].”
However, with the rush of IoT platforms entering the market, some industry watchers are paying critical attention to the direction this sector is headed, and calling for renewed focus to be paid to what an IoT platform should be expected to deliver – and to what an IoT platform actually is.
“An IoT platform is an intelligent layer that connects the things to the network and abstract applications from the things in order to enable the development of services,” says Saverio Romeo, principal analyst at Beecham Research. “[But they also] represent a very important topic for the IoT community, because they represent the ‘brain’ of an IoT solution.”
With new projects rolling out by the month, these questions of definition are ones that any organisation considering IoT adoption should be reviewing, especially as the IoT develops to form a key part of the ICT industry’s drive for extending its parameters and perimeters.
“There are myriad IoT platforms in the market – all endeavouring to be a centre of gravity for IoT implementations – offered by a range of players, from niche and industrial businesses to software vendors and equipment makers,” according to Carrie MacGillivray, VP of mobility and IoT at market intelligence firm IDC. “The challenge is that many of these platforms focus [only] on one aspect, whether device connectivity, connectivity, application management, or even analytics.” The IoT platform is receiving more scrutiny, MacGillivray continues, because it is implicitly tied to providing business value because it links the IoT endpoints to the applications and analytics needed to generate profitable business results.
Planning for IoT projects differs from standard IT implementations because for most enterprises it represents their first step towards connecting physical assets without human mediation. It also introduces a scale of complexity that will be new to many, especially when overlaid on existing ICT networks.
“Complexity is the key problem,” says Nigel Upton, senior director of IoT and MVNO at Hewlett Packard Enterprise (HPE). “Different devices have several different ways to communicate and present data to the enterprise. To reduce complexity with HPE’s Universal IoT platform, for example, we have doubled-down on making sure we can connect any type of device over any type of network. The Universal platform then translates the data into a common format [HPE has chosen the oneM2M standard].”
Usability is a prerequisite for an IoT platform to find appeal, says Matt Peddicord, VP of operations at technology startup Bug Labs. “Bug Labs’ IoT platforms are developer friendly, simple to use, and include open source code libraries, easy-to-understand APIs, free app templates, and an active user community. The entire development process, prototype to production, can occur online.”
Arrayent, meanwhile, characterises its IoT platform as an “‘Internet of Things Operating System’ in the cloud,” says the company’s head of marketing Peter Radsliff, “that offers a common set of services – e.g. firmware updates, alert engine, device management, account management, customer support, data export, ecosystem adaptors, and so on – for any connected ‘thing’ to use.”
Platform confusion
The label ‘IoT platform’ is now applied by various vendors to their products, without defining what a ‘platform’ is and/or should deliver. Sometimes ‘IoT platforms’ are ‘full stack’ solutions (i.e. enabling/managing device connectivity through to data visualisation, analytics, and APIs); sometimes the ‘platform’ might be just one element of an IoT-enabling fabric, or configured to work with some other kind of platform that undertakes the greater part of IoT functionality.
As IoT Analytics’ Scully points out, while many vendors offer an ‘IoT platform’ it can be hard to tell whether this refers to a ‘complete and mature’ IoT platform with the eight components stack (IoT Analytics proposes eight – see box-out on page 16), or whether “the term has been ‘stretched’ to describe just an element of a platform – or even something completely different”.
Cisco Jasper’s head of vertical solutions Theresa Bui concurs: “The term ‘IoT Platform’ has been quickly co-opted by everyone trying to enter the IoT solutions market, and this has caused a problem for potential enterprise customers”. This lack of a clear description has sometimes left them unsure of IoT ‘platform’ vendors, and wary of the possibility that some vendors appear to be massaging the terminology to fit the requirement.
Indeed, ‘platform’ is the most overused word associated with IoT, says PTC’s Luc Perard: “Everyone uses it to qualify anything and everything that they aim to buy, to sell, or that they report about. Consequently, many enterprises end-up comparing apples with oranges. Worse, some end-up buying a product that brings little to no value to their business. Customers are confused...”
Perard continues: “In IT contexts, a ‘platform’ is a technology base on which apps are built and run. This is perfectly applicable to IoT. Managing the connectivity of assets or IoT devices – for example, activate/deactivate service, test performance, change subscription plan, and so on – is a capability that most enterprises need. The software that delivers this capability, however, is not a platform; it is [in fact] an app.”
Tom Rebbeck, research director/digital economy at Analysys Mason, suspects that as the market matures and settles down, market forces will determine whether or not some sort of common descriptor is needed. For the time being, the details of each IoT platform differ and “direct comparisons are hard, if not impossible,” Rebbeck says. “A confusing area... But we are gradually seeing the key elements of an IoT platform emerge, and a clearer taxonomy will develop.”
Bui also wants to see “greater differentiation” between ‘platform providers’ of whatever the stripe: “[This differentiation] should demonstrate to enterprise customers exactly what problem their platform solves, and how they are making it easier to launch, manage and monetise IoT.” If this happens, many providers will abandon the generic term ‘IoT platform’, Bui reckons, and instead adopt terminology that “more descriptively helps the market understand its offering – like ‘IoT Application Enablement Platforms’, ‘IoT Service Lifecycle Platforms’, and so on.”
But Peter Radsliff at Arrayent thinks that any initiative “to limit what IoT platforms become, how they approach problem solving, or how they define themselves, will only serve to further confuse the market – and slow- down innovation.”
Building blocks of an IoT platform: the IoT Analytics model
IoT Analytics sees the IoT platforms that go beyond enabling connectivity between devices as consisting of eight building blocks:
1: Connectivity and normalisation
A connectivity layer has the function of bringing different protocols and different data formats into one ‘software’ interface.
2: Device management module.
Ensures the connected objects are working properly, and its software and applications are updated and running.
3: Database
The management of device data brings the requirements for databases to a new level.
4: Processing and action management
Data that is captured in the connectivity and normalisation module, and that is stored in the database gets brought to life in this part of the IoT platform.
5: Analytics
Many IoT use cases go beyond action management, and require complex analytics to get the most out of the IoT data stream.
6: Data visualisation
Aka ‘visual analytics’: presents an often underrated part of the IoT platform.
7. Additional tools
Development tools allow the IoT developer to prototype and test IoT use cases.
8. External interfaces
It is sometimes crucial that IoT integrates withERP systems, management tools, manufacturing execution systems, and the rest of the IT ecosystem.
Procurement pointers
One question enterprise IT leaders will ask is whether the process of purchasing an IoT platform – on a standalone basis, or as part of a larger-scale IoT solutions project – differs from other IT and communications procurement.
HPE’s Nigel Upton’s view is that the procurement process for enterprise IoT projects is not only distinct from other aspects of IT, it is also “evolving”: “When IoT started, there were many small projects, like single types of devices being connected to a single application – quick and easy to implement, and to measure ROI. As different devices got added-in over different types of networks, the projects become more expensive, more complex, with a bigger impact.”
Upton adds: “The procurement then became more rigorous, and we have seen significant rigour being applied to an IoT platform, because it is now perceived as both strategic and long term. Pricing as-a-service and on-premises are rigorously compared, and the ability to both deploy and support these ever-growing IoT ecosystems is now measured. This is why, as the platforms connect an ever-increasing number of ‘things’, there will be ever-fewer [vendors] able to invest at the appropriate levels to meet the requirements that IoT will drive.”
Bug Labs’ Matt Peddicord argues that the ‘bottom-up approach’ to IoT platform procurement is the future. Organisations can develop, test, iterate and ‘prove-out’ a new product or service with very little upfront investment, he explains: “Only when they are sure of a concept and approach does any type of traditional procurement need to take place”.
IoT platform procurement deciders
Apply some basic tests against IoT platforms’ product specs and value propositions that will help ascertain if they make a good match for an enterprise IoT programme.
- Do the IoT platform vendors havea standard definition of what a ‘platform’ is? Is this stated in its product information, and consistent wherever it appears?
- Is your IoT project to be undertaken in-house or by an external managed service provider?
- How do the IoT platforms under consideration ‘stack up’? Are their solutions ones that offer a multi-layer range of functionality? How do their component blocks or stacks compare against rival solutions?
- Clarify an IoT platform vendor’s pricing strategy, and how it scales against IoT system expansion: i.e., will a managed service cost more as your IoT grows?
- Do IoT platform vendors under consideration already have customers in the vertical sector(s) your organisation aligns with? Can they provide reference sites / endorsements?
- Try to anticipate future interoperability and integration needs based on probable scenarios; ensure that even if an IoT platform software may not be wholly open to integration with other IoT systems, the data running over it is accessible.
- Determine if IoT platforms under consideration are able to work with existing ICT estate assets, and are compatible with enterprise-critical legacy systems, protocols and operating environments.
An ‘internet of silos’?
With providers of IoT platforms tending toward proprietary configurations and not always making APIs available, organisations with a future requirement to mesh, merge, or otherwise interconnect IoT networks, or which find themselves wanting to do so because of a merger or acquisition, could find that ambition limited by a lack of built-in integration.
Bug Labs’ Matt Peddicord expects there to be a growing need to ‘merge’ different IoT platforms – “Although we would not characterise it as platforms ‘merging’. It is more about merging the data from various, disparate platforms. Again, the increasingly popularity of APIs will help facilitate this direction, and the demand for related tools.”
Analysys Mason’s Tom Rebbeck also thinks that interoperability is becoming an important factor when deciding on an IoT platform partner, especially for customers with requirements based in the built environment, where application shelf-lives must extend across decades. “This issue is certainly a concern,” he believes. “For any organisation that is looking to develop a solution to last 10 to 15 years – think smart street lights, smart metering – this inevitably will affect their choice of supplier. They will want assurance that the platform will be around for the duration.”
Is consolidation inevitable?
Rare in the history of ICT is the market sector that can provide a livelihood for 300 contenders all offering more or less the same solutions. It’s probable that some of these players’ end-game is not to grow into market leaders, but to be acquired by one of the bigger brands now entering the IoT platforms space – names like IBM, Hewlett-Packard, Oracle, SAP and Cisco.
HPE’s Nigel Upton foresees consolidation for large-scale enterprise IoT platforms happening “fairly quickly [say within] 24 to 36 months...” He continues: “The sheer scale and complexity of a full IoT platform that can connect, process massive amounts of data, perform analytics, control the devices, create micro services and integrate into existing applications and infrastructure, is huge. Consolidation will happen with the large IT companies, such as HPE, leading the way.”
Finally, interoperability between cloud ecosystems will become more important as the application specificity of platforms develops, predicts Peter Radsliff at Arrayent: “The ability to facilitate integration between those ecosystems will be the single biggest differentiator between IoT platforms,” he explains. “Those companies who delay or stumble in this approach will be the losers, and those IoT platforms who make it easy for third-party developers to make use of, and create new user experiences by tapping into, disparate cloud ecosystems, will be [the] winners.”
