Want smarter business operations? Analyze your event data

Editor's Note: In three-part package of tips, Gartner Distinguished Analyst W. Roy Schulte discusses how the right approach to event processing can make business systems not only faster, but significantly smarter as well. In Part I, Schulte defines event data and discusses methods for capturing it. Here, in Part II, he describes approaches for analyzing event data. Part III offers tips for using analyzed event data.



When it comes to boosting the IQ of your business systems, being able to continuously collect internal and external event data is a great place to start. But Gartner Distingushed Analyst Roy Schulte cautions that it's really just the first step.

The next step—and it's a doozy—is analyzing that ongoing tidal wave of information. That's where analytics come in. "There's a whole range of analytics that you can apply to this incoming data, that can help you digest it and help you understand what's happening," says Schulte, who is also a Gartner vice president. "What you're trying to do is distill the insights from the raw data that’s being collected and prepare it in a form that’s usable by a person."

Numerous techniques can be used for this purpose. "In some cases, you would be using a product like a business activity monitoring [(BAM)] platform," Schulte says. "For a little more sophisticated kind of analysis where you're trying to detect patterns, you would be using something like a complex event processing [(CEP)] engine." Other possibilities include statistical analytic tools, predictive analytics, rule engines, optimization tools and digital control systems, among other.

The main difference between those techniques and traditional business intelligence (BI)? "These analytics are continuous," Schulte says. "They run all day long as your operations are running." In contrast, traditional BI and corporate performance management are run by the clock, providing reports at certain intervals. "Maybe it's every hour, maybe every day, maybe every month," Schulte says. "Or they’re interactive, where you ask the system to give you a report now—but if you don't ask, you don't get the report."

In real-time operational intelligence, though, the system is constantly running and processing events. "When new events come in, it's continuously re-computing to figure out what's happening and what should we do about it," Schulte says.

Two objectives: visibility and 'situation awareness'

"The most common thing I hear from people considering this kind of computing is that they're looking for visibility," Schulte says. In other words, businesspeople want to see what's happening both inside and outside their companies in near-real time.

Others are seeking "situation awareness," a term rooted in the command-and-control military structure. "In the military, what they wanted to do was cut the fog of war," he says. "In business, we have a similar fog. We have the fog of commerce, where there are so many things going on." Businesses need to cut through that fog to understand what's happening, he says. They need to create a 360-degree view that puts business activity in context—but without becoming overwhelmed by the sheer volume of raw data.

Where is all that data coming from? "It may be data from process monitors, telling you how the process is running, how the individual instances of the process are proceeding," Schulte says. "Various kinds of key performance indicators or key risk indicators may be being displayed. You may see classic reports like bar charts, pie charts, line plots, scatter plots, graphical displays and so forth."

Just a decade ago, such real-time information was rare. "You would see it only in a few places like in a network control center or perhaps in the military, but you didn't see it in routine business operations," Schulte recalls. Now dashboards or consoles can provide business users with that all-important situation awareness, and most new application systems have at least a couple of continuously updated displays monitoring various part of the operation.

Schulte cautions that, in most cases, businesses aren't yet getting true 360-degree situation awareness. But it's a place to start. From there, they can focus on expanding the amount of available data while, at the same time, making sure users don't find themselves drowning in information they don't need. "You really need to focus the design of the systems on the concept of management by exception, where you're showing people only the data that they need to have to make a decision to take an action that they need to take," Schulte says. "We're talking here about dashboards that aren't just displaying information. We're talking about dashboards that help you actually take an action without having to switch to another application."

In Part I of this package, Schulte defines event data and discusses ways to capture it. In Part III, he talks about how to take action following event-data analysis.

About the Author

Anne Stuart, ebizQ's editor from mid-2010 to mid-2013, is now senior editor for SearchCloudApplications.com at ebizQ's parent company, TechTarget. She is a veteran journalist who has written for national magazines, daily newspapers, an international news service and many Web sites. She’s specialized in covering business and technology issues for 20 years. Based in Newton, Mass., she can be reached at astuart@techtarget.com. Follow Anne on Google+ and at annestuart_TT on Twitter. For general questions about ebizQ, please e-mail editor@ebizQ.net.

More by Anne Stuart, Contributing Editor