I used to hear the phrase "location, location, location" in real estate ads. If enterprise use cases for IoT are any indication, that phrase might just as well apply to IoT in the early going.

"The key use cases for IoT right now are in real-time location tracking," said Steve Wilkes, founder and CTO of Striim, which provides real-time data integration and streaming analytics software.

Wilkes cites the following common use cases:

  • Airports and hospitals are using IoT sensors to keep tabs on wheelchairs and their locations so they can track and find wheelchairs when they need them and ensure optimal use.
  • The use of personal healthcare devices is exploding in the market, with one of the latest innovations being a new version of the Apple digital watch that automatically tracks your blood glucose levels. Especially useful for users with diabetes and other blood sugar concerns, devices like these provide real-time feedback. There is also the potential to collect, anonymize, and aggregate this data so healthcare institutions can perform analytics to provide preventive and proactive healthcare.
  • Logistics companies are sticking sensors on almost every box, container, pallet, and package to ensure the end-to-end tracking of goods from point of origin to final destination as goods flow through the supply chain.
  • Airport and mall managers are beginning to track foot traffic so they can assign appropriate rental rates to different storefronts on their premises based upon foot traffic patterns and volumes.

What's positive about these use cases is that they are easy to cost-justify to the business and relatively easy to deploy. However, in the longer term view, IT still has the job of managing all this data that is streaming in from data points in geographical distribution networks that can grow quite large.

"Trying to implement IoT as a separate and siloed application is a bad approach," Wilkes said. "And it is tempting for companies to do this because most have an antiquated information architecture."

These architectures were originally constructed around data center assets and systems of record that were all neatly organized into fixed-length records that could easily be stored and queried.

That isn't the case with IoT, where data is unstructured, and where there are many data types that don't necessarily work well with each other. The plot thickens further when this unstructured data streaming in from IoT must be meshed with the highly structured, fixed-length records of data center systems.


featureddownload2.jpg

Hiring kit: Data architect

The data architect develops the foundational architecture of all data generated within an enterprise--an increasingly critical, challenging position to fill. This kit includes a detailed job description, sample interview questions, and a basic want ad to simplify the task of finding the best candidates.
Free for Tech Pro Research subscribers.


"Getting all this disparate data to work together is a challenge," Wilkes said. "This is why when enterprises deploy IoT, they should look at all the different IoT devices they are deploying at the edges of their networks to see if there are communications protocols that enable these devices to transmit their information and to exchange with each other over the internet. Enterprises should also be deciding what data these devices transmit that they really need--and what they don't need."

In many cases, devices are exchanging "handshakes" with each other to connect, but this has no inherent data value for the end business. If computing placed at the edge of the enterprise can preprocess and exclude this data, there will be less data to transport and store in central databases.

"Enterprises also need to look at their data lakes and other data repositories so they can determine which of the real-time IoT data they want to store and even aggregate for query and analytics purposes," Wilkes said.

Developing an end-to-end architecture that addresses both centralized enterprise computing and IoT is in early stages at most companies. It is an area that needs to be addressed by data architects.

"This is important because there is a big difference between having data and having information," Wilkes said. "For instance, you can slap sensors on motors and couple them with smart power supplies to get information on usage, but what if the temperature that the engine is operating at is too high or there's too much vibration? Do you have an image of the engine you can look at? All of these elements need to be brought together so you can look at the data in different ways and get the complete picture."

So how do companies move beyond elementary IoT locational data reporting and into more value-added information?

  • Begin by assessing the communications capabilities of the IoT devices you are deploying at the edge of your enterprise. Can these devices communicate over the internet to an on-premises or cloud-based system?
  • Revisit your network infrastructure. Do you need to place more computing power at the edge of your enterprise? What types of storage, security, and processing are you likely to need?
  • Review your enterprise data architecture. With IoT rapidly adding to your data under management, which IoT data do you want to discard? For the data you save, where should it go? To the cloud? To an on-premises data repository? To a local edge server?
  • Which IoT data do you potentially want to aggregate and/or optimize for use in analytics data marts, and for what business purposes?

"There are many approaches to IoT that we talk about, but in the end, enterprises need to be thinking about what they want IoT to do in real time for them at the edge--and how they expect it to add value later for them in data lakes and repositories," Wilkes said. "The best way for companies to do this is to take a step back and think about their current IoT use cases and then create a strategy that links this IoT data to their analytics and central resources so they can further build business value."