Print This Post

Challenge of Unique Data for the Internet of Things

Originally published on LinkedIn:

Illustration showing how silos of device data are connected through master data to serve many business purposes

Everyone is familiar with the classic case of master data management (MDM) gone awry in which a bank, having just sold a mortgage to a person, immediately sends that person an invitation to buy a mortgage. The reason this happens is that there are two databases — one is a database of potential customers that is maintained by marketing, and one is a database of actual customers maintained by the mortgage lending department — and when the person takes out the mortgage, they are not removed from the database of potential customers when they are added to the database of actual customers. Such examples are numerous, and arise whenever multiple organizations or departments are involved in independently managing information about common core business assets such as products (like mortgages) and customers.

Similar problems can be expected in the land of the Internet of Things, and for much the same reasons. If there are a great many devices deployed across an enterprise, it is quite likely that different interests in these devices will lead to many different databases and probably also to many different identifiers and descriptions, and the same MDM problems will arise. A sensor that has just been installed may already be scheduled for a maintenance inspection, or even for replacement, or when an application fails we are unable to identify the devices that may have led to the failure.

Due to the large number of IoT devices and device types, we can anticipate many different corporate functions to be involved in managing them. How devices are managed may be organized by function, by department, or both — for example: 1) application design, 2) installation, 3) calibration & testing, 4) maintenance, 5) operations, and 6) leasing. In addition, an IoT device may be explicitly visible or it may be hidden inside some piece of equipment or machinery.

These and many other problems can be expected to arise because we do not have unique data about a device, or unique identification of the device across the enterprise, either functionally or from one department to another. Multiple databases are likely to be created at the department and functional level with each database containing different, and typically non-synchronized, information about the same device. Moreover, this lack of synchronicity will likely extend to the description of the real world objects with which the devices interact, even in cases where a common databases of devices actually does exist.

The resulting problems can cost large amounts of money directly, and can cost even more through collateral impacts like the shutdown of a plant or a critical plant operation.

Master data in the traditional sense focuses on the key, and relatively static, data of the business (such as customers and products) and ignores the day-to-day business transactions (such as sales). In the same way, the focus for IoT that we are talking about is on the key business metadata concerning IoT devices, such as device identification, device description and properties, device connectivity, location context, and application context (which identifies what application(s) use the given device).

Mastering key metadata for IoT is fundamental to IoT success, and even more so to the longevity of IoT applications.