Predictions and Payoffs - The Story of Operations Analysis

Big Data Product Marketing Manager, IBM

1.21 Gigawatts?!?!

Here’s a big data problem for you. Let’s say you’ve accidentally traveled back in time 30 years and the only way to get Back to the Future is to transfer 1.21 Gigawatts of energy into a beat-up DeLorean.BackToFuture.jpg

Well, back in 1985, the solution, by Hollywood standards of course, was the clock tower. If you knew when and where lightning would strike, then in the words of Doc Brown, “it just might work.”

Nearly 30 years later, maybe we haven’t mastered time travel, but we’ve come pretty far with predicting outcomes. In fact, that’s a key issue at the heart of one of the five high-value big data use cases: Operations Analysis.

What is Operations Analysis?

At a basic level, it’s about analyzing a variety of operational data, including machine data, to improve business results and decision making. But what does that actually mean? Organizations are overloaded with data; we all know that. But operational and machine data in particular can be very challenging because of the sheer complexity and massive volumes of it. It tends to come in various formats, making it difficult to analyze across data sets, and it grows exponentially.

Yet those who are able to tap into this data can start to do some pretty cool things with real-time monitoring, gaining visibility into operations, transactions, customer experience and behavior as it’s unfolding, and proactively planning to avoid inefficiencies, rather than react after the fact.

Predictions, power lines and payoffs

I have to admit, the ability to predict the future was always a bit of a fuzzy concept for me. How can anybody know what’s going to happen and when? How realistic is it to prevent an outage, for example? A conversation with an executive at an energy company about power lines a few years back made things clearer.

Picture a power line with a tall, old tree nearby. Over time, one of the tree branches starts to droop and begins to hit the line. The taps are light and infrequent at first, but as time goes on, the branch begins to press on the power line more and more. Eventually it snaps, cutting the power line, putting hundreds of customers in the dark, and sending the energy company into an emergency situation to restore the downed line.

Now consider that same power line, but with sensors all along it. These sensors detect motion, in real-time, so that when the tapping begins, it alerts the energy company that something unusual is afoot. This time, the energy company can send a crew out where the tapping is occurring, see what this low branch is doing and remove it before it snaps. Could the energy company have been able to guess on their own the one branch on the one tree that was destined to fall? Never. But the simple notion of collecting sensor data allowed them not only to know it was coming, but prevent it from happening.

That’s what operations analysis is all about: the ability to work with operational data, even in real-time, combine it with your enterprise data, and visualize it in a way where you can put it directly in the hands of the decision-makers, who can then call for the ominous tree branches to be cut.

Tune into this podcast on the operations analysis use case to find out how other industries are mining machine data to improve efficiencies, what kinds of technologies are required and how other organizations have overcome the inherent challenges of leveraging machine data to open up a new world of possibilities.

Other useful resources

Learn more about all five of the top high-value use cases for big data