Why are real-world IoT implementations still so challenging? Because there is a huge amount of complexity in accessing data from disparate endpoints. Getting data from the edge—whether from a robotic manufacturing arm in a remote location, a smart vehicle or a shipping container—is incredibly hard.
It takes some businesses years to build out the infrastructure needed to connect to their remote assets. Seemingly simple deployments hide major challenges that can drastically delay projects before developers can write a single line of code. And there are many questions to answer to ensure you can get the data you need.
Building the infrastructure to bring asset data into your enterprise—securely, reliably, at scale—requires expert knowledge of widely diverse topics On the IoT technology side alone, you need extensive skills sets in:
IoT applications also present unique business challenges, including:
ABI Research, “IoT Market Tracker, Q1 2018”. Gartner, “Predicts 2016: Rising to the Challenge of Building IoT Solutions”. 24 November 2015. Benoit J. Lheureux et al.
Few organizations have both the broad base of technical know-how and the experience operating IoT systems at scale to succeed with their first IoT implementation. It’s the biggest reason why nearly three quarters of IoT initiatives fail.
Many enterprises now recognize just how complex IoT implementations can be and, increasingly, turn to partners that can bring the requisite skills. But even if your IoT partner does an excellent job building out your edge connectivity, you’re still only part of the way there. After all, IoT is not about infrastructure. It’s about data and the insights gleaned from those data. If you want to get meaningful insights—and do it efficiently, in a way will scale with hundreds or thousands of real-world devices—you’ll need to think very carefully about how you’re handling data throughout the stack. You’ll find two big problems here to address.
Most enterprises are not prepared for the avalanche of data that thousands—or hundreds of thousands, or millions—of assets will generate. They’re likely to find themselves buried in data coming in far too quickly to analyze, process and act on. What’s needed is scalable IoT data orchestration. Enterprises need a way to preprocess all data flows coming in before the application ever sees them. That preprocessing should live not just in the cloud, but at the edge, and all the way down to the deep edge device embedded in the asset. And, if organizations want to be able to quickly adapt IoT applications to changing technology and business requirements, they should be able to refine and rewrite code using the same software tools their developers use now.
The second big problem: making sure that your IoT application doesn’t render your deployed devices useless by burning through their available power too quickly. New LPWA technologies give you a much more power-efficient framework for data collection than legacy technologies. But if you’re not using them correctly, they won’t last anywhere near the lifetime that many applications require. For IoT devices, transmitting data consumes power. If your devices are going to operate for years in the field, they need to transmit data intelligently—sending only what you need and only when you need it. The problem is, there’s a good chance you won’t know the optimal formula for shaping your data transmissions until after you deploy. It’s quite common, for example, to realize that some data for which you thought you needed a daily or hourly reading is much less volatile than you expected—or that other readings you’re measuring should be transmitted more frequently. If the logic dictating how and when your devices transmit data is hardcoded into their processors, you won’t be able to make those kinds of changes easily. What you need is stream processing intelligence at the edge. Ideally, you should be able to run data management logic on the modem itself in a lightweight and flexible way. And your developers should be able to deploy changes to that logic at any time, in real time.
IDC, “Data Age 2025”.
Organizations deploying IoT applications have always struggled to overcome data management and transmission constraints. But the existing options leave significant complexity—and significant gaps—for organizations to wrestle with largely on their own. Current approaches fall into two basic categories:
