I recently read a summary of an article called “Analytics for the Industrial Internet of Things: Not Your Father’s BI” by David White of ARC (available to ARC’s clients). He says that “business intelligence and analytics are due for a revolution” if they are to support industrial IoT applications.

Real-time events

The use of the business intelligence (BI) lens through which to view the IoT problem is interesting. This perspective is common. People who have a history in BI, and often more accurately in data warehousing, have years invested in a paradigm that has historically served up information in order for humans to detect the conditions that justify action. For those of us who have been brought up seeing the world as a series of real time events (rather than a repository of historic events) we wonder why one would wait for a report to act on a condition whose significance was understood prior to the event.

Time for a different tool

David White makes the same observation but insists that BI must change. That sounds like insisting that a roofing hammer must become a rubber mallet for use on furniture. The tool does not need to change, a different tool should be used. 

Part of the "change" is the adoption of complex event processing. This wording is the problem. It is  "data warehousing" people who think complex event processing is a "change." Data warehousers have co-opted the BI term as a means to keep the marketing fresh on their backwards-looking solutions.   

A BI paradigm shift

If you ignore the tone of White's comments and view them through the lens of event management and correlation, his prediction of what your organization will need is consistent with our view. 

When it comes to BI, the Industrial Internet of Things will shift emphasis to:

  • A Real-time Focus – BI has traditionally focused on the analysis of historical data. “Data analysis may occur as soon as it is captured, while the data is still ‘in-flight.’” Complex event processing engines become a core technology for analyzing multiple data streams simultaneously and generating alerts and alarms. Higher level predictive and prescriptive analytics will create larger frameworks that will allow diverse events to be correlated, quickly interpreted, and then acted upon in a timely manner.

  • A Rethinking of Data Storage Practices – The vast majority of IoT data will have only fleeting value. But companies need to think through this carefully and determine what data should be stored to help drive continuous improvement and what data can be almost instantly disposed of.

  • Business cases for determining the ROI of IoT Analytics – the business cases for IoT analytics will be very different than that of traditional BI.

White goes on to say that “collecting, storing and analyzing IoT data requires different processes, skills and technologies.” He is right ... if you are still thinking in terms of data warehousing. If you are building the event-based organization it won't be nearly so different. 

Where are you in the conversion from report oriented management to event oriented management? 

Michael Lee is the Managing Director of Quality Deployment. He can be reached at insights@qdbve.com.

The Best Kept Secret for IT Professionals.

Learn how we keep you more informed than your competition in half the time.

arrow

No Comments

Add a comment