What EAs Need to Know About…the Internet of Things -

An Interview with Kary Främling

By Association of Enterprise Architects

We recently introduced a new series on the Association of Enterprise Architects’ blog called "What EAs Need to Know About…” that looks at industry trends and issues that Enterprise Architects need to consider as they put together architectures and do their day-to-day work.

In the second installment of this series, we talk to Kary Främling—Chairman of The Open Group’s Internet of Things Work Group, CEO of ControlThings and Professor of Practice in Building Information Modeling at Aalto University in Finland—about the things Enterprise Architects need to consider as the Internet of Things begins to become a more prominent part of enterprise technology strategies.

The Internet of Things (IoT) is one of the hugest trends to hit the technology industry in years. How do you see the IoT affecting the discipline of Enterprise Architecture?

That’s a big question. It definitely means a lot of change for traditional information systems—enterprise systems made for managing operations of companies, orders and sales. What is happening now is a lot of companies producing any kind of goods or machines would like those machines to also communicate with their own information systems and other systems. The biggest challenge with this is that, in the past, organizations might have needed to communicate just with other organizations. Now, if they need to connect devices or sensors from the Internet of Things to their products or systems there might be hundreds of thousands of sensors and also other information systems to connect to, which may or may not be enterprise systems. They might be installed in homes or cars or so on. That makes the whole information system ecosystem or landscape look quite different. When you consider the whole picture, you will have loads of different organizations and people and mobile devices that you need to communicate with. That really is a scalability issue. It’s also a challenge from a security point of view. Many people call this new information system architecture a "System of Systems.”

Will Enterprise Architects in some industries be more affected than others? Which industries can expect to be most affected?

The companies that are really "all in” are the companies that are making complex products, such as General Electric and their "Industrial Internet” concept. Here in Finland, for example, there are companies like Kone, which makes elevators and escalators, as well as Konecranes that makes large cranes for harbors and other places. There’s also Wärtsilä, which designs and produces diesel engines. These companies are now trying to get visibility into their equipment, in similar ways as Rolls Royce has been doing with their jet engines for many years. Rolls Royce is able to monitor usage hours and maintenance information of engines and collect that data in near real-time. It is even being said that Rolls Royce engine data was the last information received from the Malaysian Airlines plane that disappeared in 2014.

The key word is "servitization,” and the need for product services across the lifecycle. For example, manufacturers have to modify their business to reflect more of a service model. They have to implement this visibility into their equipment and will have to somehow be able to collect that information and communicate with products even though they may be in different places or buildings or organizations that they don’t regularly have contact with. That’s really the place where I see the Internet of Things breaking through seriously.

Our company, ControlThings, has made a system for the air handling units of building HVAC systems that is really an example of the next phase of building automation. The system can be controlled via a smartphone once turned on inside a building. It provides alerts to facilities management regarding performance and when there are problems. This is happening rapidly in building and facilities management, providing visibility into how buildings are performing and providing service operations for them.

This is also coming in vehicles. There are already systems like this in many trucking operations, and Tesla has Internet connectivity. You see this a lot in high-end, expensive vehicles, but those still tend to be closed systems to a certain extent.

The next wave will come from vehicles and then slowly make its way toward consumers and consumer products. It will start in industry and then move slowly to consumer systems, vehicles, houses, buildings and so on.

Hospitals are another area. We have been working on this, but for some reason it seems to be going very slowly, perhaps because the hospital environment is so regulated and closed. One place where we also see this coming, in the U.S. in particular, is with insurance companies that are ready to give discounts to people who agree on system measures for exercise and health. Once people get used to seeing what buildings are doing and that their houses are smarter, and that they can communicate with them, they might be more comfortable with providing health access, but personally I don’t see that having wide adoption for a few years.

For Enterprise Architects in industries that will be heavily affected by the Internet of Things, what will they need to prepare for?

The usual answer is that the amount of data that will be collected will just explode. Of course you do get these questions of how do you handle Big Data. We have made some studies in the past with our platforms about whether they should start with high resiliency computers on cloud servers or having hard disks of their own. It’s a challenging question. We have noticed that cloud storage could be extremely expensive for the amounts of data that we have.

It is a new challenge for Enterprise Architects, and the amount and type of data is more difficult than traditional data.

Are there particular system design problems specific to IoT environments that should be considered when planning an architecture that will include the IoT?

Yes. Actually the current view is that you should just collect data on your own systems to do your own analysis. That’s not the vision of the Internet of Things. The IoT should be about having different systems communicate. For example, to optimize energy consumption in your house, you communicate with the energy provider, your house and device manufacturers. That has to be a much more interactive information system to be able to retrieve data from machines and different data sources and all in real-time. That’s a problem to figure out how to treat that data.

I’m quite convinced there will be more and more intelligence embedded in products themselves, and there will be more intelligence at the edges of the Internet of Things. Companies will want to be able to react to incoming data and possibly perform corrective actions directly.

The current vision many people have of the Internet of Things is to pile up a load of data collected about the products and then eventually throw those terabytes of data into an engine for analysis so that you can come up with something useful. We have seen that sometimes comes to the point where the data isn’t useful, and you’ve forgotten why you collected it in the first place.

Data needs to be context dependent. For example, if a check engine light comes on in your car and you collect the sensor data for the following half hour and the previous two days, that’s extremely valuable data to collect, but it’s context dependent. Data from the IoT should not be piled into back-end systems or databases without thinking about how you’re doing it. You have to take the information about the products and people using them into context, taking into account by whom and when the product is used —the systems information has to be relevant.

Is much of architecting for the (IoT) going to be about having a well-architected Data Architecture?

Definitely. But it’s not just the data architecture. It’s also about moving intelligence toward the edges of the Internet and the ability to react on the available data. You should have data archived that allows you to analyze it and extract new intelligence or knowledge that you can embed into the products so you can improve their performance, as well as their design and manufacturing.

What are some of the ways that Enterprise Architects should go about incorporating the IoT into their architectures? How can they start?

What all these information systems are moving toward is REST-based interfaces. The Open Messaging Interface (O-MI) and Open Data Format (O-DF) Standards would be important for Enterprise Architects to have a close look at and use them straight away from the beginning because it’s not just about what data organizations are collecting anymore. It’s about what you might have to communicate with in terms of thousands of devices (or much more) and possibly hundreds of other organizations. If you don’t start by using some limited set of standards, the whole architecture might be really hard to manage.

Are there different considerations Enterprise Architects need to make when incorporating such a large network of systems into their architecture?

Security, obviously, but that’s too simple to say just that. It does pose a very large challenge in how to manage authentication of users and other information systems. It will require much more precise control of what information you provide to different users or systems. The granularity of access to different systems is much more challenging than it used to be.

For instance, with my car, the information that I would be ready to provide the local authorities in Finland is not the same information I would give to my maintenance provider, and that’s not the same information I would give to the manufacturer or to my wife. There’s also the concept of role-based access to different information.

Firewall configurations are also a concern due to the need for even more fine-grained access models when it comes to roles and information items. It is already now a challenging task for IT engineers to configure firewalls properly, i.e. providing selective access to specific interfaces and data.

The number of different systems will grow radically in addition to the aspect of user roles and the granularity of information. Those are huge challenges.

What would you say are the biggest challenges in architecting for the IoT?

I think it is crucial to address those security issues properly. Also, the scalability of the whole thing, the "System of Systems.” Another thing that people have started talking about—and they’re trying to tackle this—is the different roles of people and different devices.

For instance, if you want to make a really secure system, then you would probably like to use or have a SSL layer. SSL currently relies on usage of certificates issued by certificate providers that make of up the Public Key Infrastructure (PKI). They work for sites such as banks or companies of all kinds, but it quickly becomes a scalability issue of managing certificates with the current PKI. That might be a challenge when you go to the level of certificates for products or different people so you know for sure when I pretend to be me in the role of the owner of my car that I really am me. How does my car know and trust that it is me using the system?

What will some of the benefits of the IoT be for Enterprise Architects?

Many of the main benefits will not be for architects themselves but from the new business models and the money companies can make from adopting IoT concepts and technology. Some companies are also creating a new position called the Chief Digitalization Officer that seems to sit between the CIO and Enterprise Architects. They define how to make products ready for the Industrial Internet.

I get very enthusiastic about the possibilities of new architectures. The IoT really opens up possibilities for new, more "intelligent” products and systems. I believe that this is a major reason why I see our students at Aalto University and others becoming more interested in Artificial Intelligence again. People are seeing these systems coming closer to reality and what people are doing with their lives.

For Enterprise Architects, things should become more fun. Look at self-driving cars and the boosts in computer science all over the world. Enterprise Architecture will be part of that, and a lot of Enterprise Architecture will become something new.

I don’t think there will be many companies or industries that will avoid being touched by the IoT. Even banks will eventually be affected by it—all of this will be connected to the ecosystem of mobile phones, who will communicate with the environment and the need for supporting micropayments. If that comes, even banks will have to modify their systems architectures. I can’t think of any industries where this will not eventually affect Enterprise Architects.

Are there some early adopters who are already architecting well for the IoT?

I think there are many. Cisco has been thinking about producing an IoT platform and Ericsson has also been thinking about this for quite a while. IBM, Microsoft, Google—it’s mostly provider companies.

What I have seen in those IoT architectures is they are still very focused on collecting data into vertical applications, without taking context into account. They still have some work to do on that. We’re still missing a lot of the horizontal interoperability capabilities needed for retrieving all the needed contextual information. What we need to really understand is how to enable architectures that would not just be vertical.

With increasing horizontal interoperability, I’m quite convinced there will be possibilities for new ecosystems of small companies who start collecting information and create new applications—location dependent or otherwise context dependent—as well as for the people creating those information systems. EA in those companies might look quite different from the traditional ones. I’m not sure what those would look like but they’d require breaking down some silos. I expect that we will see such ecosystems emerging within the next five years or so.