Avoiding pitfalls in IoT integration - Gartner

Internet of Things (IoT) projects can involve a large number of devices, a proliferation of application programming interfaces (APIs), and large amounts of data — all of which can tax the traditional strategies, skills and technologies used by integration professionals.

Gartner research vice president Benoit Lheureux says "Through 2018, half the cost of implementing IoT solutions will be spent integrating various IoT components with each other and back-end systems.

"It is vital to understand integration is a crucial IoT competency," says Lheurex.

He says there are five steps for integration leaders that can address some of the most common challenges in IoT integration.

The five steps are:

Adopt an API-First Approach An API-first strategy is particularly relevant to IoT projects because they rely heavily on mobile and cloud computing, technologies which already use an API-centric approach.

This should not, however, be misinterpreted as an API-only approach.

APIs alone do not sufficiently address all the capabilities needed to securely and reliably scale up integration in large distributed systems.

"Determine the integration requirements for a specific IoT solution that can't be handled via APIs, and then assess whether the built-in integration capabilities of your IoT platform will suffice," says Lheureux. Identify Communication Requirements for IoT Devices First, identify how 'things' in your project will communicate, and select the best technology accordingly, ranging from cellular networks to short range wireless such as Bluetooth or ZigBee.

It's also important to consider factors such as the number, and type, of things and how different technologies will handle these variables.

Then look for the network topology — considering the potential role for edge computing or gateways — that best suits the specific requirements for device autonomy, localised computing, device aggregation and so on.

Once these areas have been qualified, it's possible to assess whether a bundled IoT platform meets requirements for the project, or whether extra solutions will be needed to build a network for things. Leverage Cloud for Data and Process Integration This step focuses on integrating IoT platforms with core business processes.

Most Gartner clients report that the built-in integration capabilities of their IoT platform are good enough for initial deployments, such as an exploratory Mode 2 project.

"Consider using your IoT platform for the initial implementation and then use a commercial integration solution, like an iPaaS platform, to scale up the project or support more complex integration, to implement workflow, or multiple IoT projects, or to access advanced integration features such as high-performance, general-purpose translation," Lheureux says. Selectively Use Traditional Software Most mid-to-large-sized organisations have substantial investments in traditional on-premises integration middleware.

While these tools are generally not optimised for IoT device connectivity or cloud services integration, they can likely help if your project is on-premises, or if your IoT platform must integrate with data and applications that are mostly on-premises. Use API Management Tools API management capabilities vary widely in different IoT platforms or other types of middleware.

At the minimum, you should plan for the possibility of adding a third-party API management solution to your IoT project, to ensure secure and reliable scaling as APIs proliferate.

This is especially true if your project involves many APIs, or has APIs that connect with many consumers, are exposed on public networks, or return sensitive or restricted data.

"IoT projects are often associated with disruptive business models and in the excitement, an appreciation of the potential complexity can be lost," says Lheureux.

"To avoid disappointment, integration leaders must address the diverse integration requirements for IoT projects.