IT, OT, and vying for control of the IoT connectivity platform
Internet of Things (IoT) “platform” has become a term as nebulous as “cloud” was a decade ago, partly because the technologies are still taking shape. However, there are clear differences between an IoT connectivity platform that serves industrial markets and those for the consumer IoT, such as reliability, security, and integration with highly specific enterprise services like building and plant management systems. But since Industrial IoT (IIoT) platforms touch so many parts of both the operational technology (OT) on the plant floor as well as backend IT operations, who should control them? Mark Benson, CTO of IoT software platform vendor Exosite gives his opinion.
There are an untold number of IoT “platforms” on the market today. What do businesses evaluating an IoT platform need to consider to make the right choice?
BENSON: It’s a bit of the Wild West for IoT platforms right now. There are a lot of differences between them, so it’s important to take this question seriously because there are a lot of buzzwords flying around as the class of enterprise software platforms for the IoT continues to emerge.
One is to look for a platform that is able to integrate operational technology and engineering with IT and the business – connecting machine data with enterprise IT is key. This new class of enterprise software platform needs to have features for connecting with embedded devices securely with protocol support that works for the application; features for managing device provisioning, authentication, and authorization, not only of users but also of devices; and also the ability to create user experiences from that authentication and authorization.
Another thing to look for is a company that really has features that make it easy to create Internet-connected products. Not just the building blocks of data bases and Linux instances for creating servers, but more of an integrated platform that brings that together in a way that provides interfaces for the OT stakeholders – the engineers that understand the machines – so they can manage their connected products, but also interfaces for the IT stakeholders so they can manage integration with their enterprise environments.
Finally, a company that has access to professional services expertise and the ability to provide guidance or best practices for successful deployments.
Those are all areas organizations should look at when evaluating an IoT platform.
Is there a difference between enterprise platforms that support Industrial Internet/Industry 4.0 deployments and those that serve the general IoT industry?
BENSON: Platforms that are looking to connect consumer products don’t have as many features that revolve around enterprise IT security or policy management for business-to-business (B2B) operations.
An example is an industrial original equipment manufacturer (OEM) that makes pumps. Their customer is a company like a waste management company that might integrate that pump into a waste truck, or into a crane, or a piece of heavy machinery, or onto an oil platform. Those customers need to be able to extract value from that pump and how it is integrated into the larger system before it can be used by a technician, a mechanic, or an operator of the larger industrial machine. The number of different users that need to interact with the platform is much more complex with an industrial-focused platform that with a consumer platform, which is really just used to help create a connected toaster or another connected devices that has a very clear user persona – it’s just the end user.
Areas of difference with an IIoT platform are its focus on reliability, scalability, security, and the user experiences and integrated workflow that make sense for a B2B type of product, where a product goes inside another product that then gets integrated with the final system. A platform that has those layers of user interfaces and access for those stakeholders are critical, in addition to integrating with specific external services. With a consumer IoT platform you’d often want to integrate with a customer relationship management (CRM) tool or something that gives you insight into how your products are used or deployed, but with IIoT applications there are integrations with plant management systems, building management systems, and maybe SAP so you can dispatch a technician to fix problems in the field, in addition to support interfaces and manufacturing tools. The emphasis on high integration with enterprise IT is really critical for industrial applications, so the platform you’re looking for needs to have those types of features.
Given the difference between IT developers and operational technologists/engineers, who would be responsible for the management of an IIoT platform?
BENSON: It’s a mixture, but it’s also shifting. Today it’s mostly driven by OT engineers that want to make a connected version of their product by adding wireless capability or a gateway that can send data to the cloud, and they’re starting to learn about the enterprise/IT world. But that is shifting into the enterprise/IT world as we speak, as enterprise/IT organizations are starting to understand that the IoT is really going to affect them, that their company’s products are starting to get connected and there may be security risks associated with doing that, etc. So the primary buyer of IoT platforms is shifting from OT to IT, and I would expect that shift to continue over the next two years.
Long term, the future buying decisions of IoT platforms will actually be governed more by marketing teams and how they’re monetizing data, getting better insight about their users, and so forth, but we’re not there yet. That’s more in the 5-10-year timeframe, I would estimate.
The reality of the platforms today is that they need to allow both OT and IT stakeholders to manage the platform, which is why at Exosite we built specific interfaces into the Murano platform that allows engineers to manage connected products, as well as specific interfaces for managing business and integration with enterprise IT as a piece of the enterprise portfolio of software (Figure 1). So it’s both, but it’s also shifting towards IT.
But don’t the basic processes and procedures of Industry 4.0 necessitate someone on the ground with intimate knowledge of the physical systems being responsible for managing the IoT platform in some regards to make sure that the wheels keep turning?
BENSON: Although the buying decision is shifting, the user experience and interfaces will always have to be tailored to OT since there will always be a need for engineers to manage and see data about their products in an IoT platform. The IoT is a movement towards data and using data to do more with less. None of that is possible unless you have deep insight into the machine and how it operates, and that is squarely on the OT side of the house with the engineers and the people who are actually building the machines. It’s critical that even as the buying decision may shift over time that OT team members have direct access and the ability to manage those connected products.
In the future it will become easier to create web applications without deep knowledge of the system itself, but it’s a pipe dream to think that those two worlds will be decoupled completely. To create really valuable user experiences and applications requires deep insight into the product itself and how it’s being used, not simply through application programming interfaces (APIs) but through access to data and understanding the context of that data and how it’s being produced and how it’s being used.
Moving forward, will there be a requirement for data scientists and analysts to be on staff at every reasonably sized OEM in the world?
BENSON: That’s the inevitable outcome. Companies that succeed are going to need to have people on staff that know how to analyze data, but I think it’s a chicken-and-egg problem. It’s a maturity model. The road to advanced analytics starts with basic connectivity.
I’ve seen companies hire data scientists to try and figure out what’s going on with their data, but they haven’t captured enough data for the data scientist to actually do anything, and the data scientists don’t understand the machines well enough to know how to instrument them to get the answers that they want. So it turns into this cycle of inefficiency where if companies don’t get the first part right – which is basic connectivity, adding sensors to machines, getting data flowing, and learning to collect data on failures – if they don’t do that first they won’t ever be successful adding data scientists to their teams and driving true operational efficiency.
I do think that is the inevitable outcome for reasonably sized industrial OEMs, but it just depends where they are in their maturity scale whether their ready for it yet or if its still out in the future.