Gartner: Bimodal IT needed to navigate centralized/decentralized debate

Gartner urges CIOs to craft a more nuanced IT strategy that is both highly standardized and highly flexible: The Data Mill reports from the Gartner Business Intelligence and Analytics Summit.

Digital business initiatives are colliding with legacy IT systems and, because businesses have a foot in each world,...

it's forcing CIOs to face new dilemmas. Should their IT strategies focus on standardization and keeping on the lights? Or on providing the kind of flexibility needed for experimentation? Neither one is the correct choice because, as CIOs are well aware, they are increasingly expected to provide both.

At the recent Gartner Business Intelligence and Analytics Summit, Gartner analysts told participants that successfully navigating this dilemma is possible, but it will require some CIOs to retool (and reimagine) their current IT strategy for self-service business intelligence (BI). Take the centralized IT versus decentralized IT dilemma. Rather than get stuck in either/or thinking, Gartner analysts encouraged CIOs to starting thinking and/and by building a bimodal IT strategy.

Back in the day, when CIOs introduced a single, centralized BI team to their organizations, the strategy was completely logical. The centralized team was established to ensure the information used to make business decisions was consistent, according to Kurt Schlegel, an analyst at Gartner Inc.

But, as the demand for data and reports increased, the centralized team became a bottleneck. "The lack of responsiveness changed the rallying cry from 'single version of the truth' to 'self-service,'" Schlegel said during the summit's opening keynote. The lines of business not only wanted access to data, but they also wanted to analyze data and build their own reports. New technologies, most notably the cloud, enabled the lines of business to do just that, creating what many CIOs refer to as shadow IT.

But, Schlegel said, rather than cast a negative light on the phenomenon, perhaps it's time to think of it differently -- as citizen development. "Such citizens are informal champions and they make stuff happen," Schlegel said. It's not that he's advocating for centralized IT to disperse; he's, instead, advocating that CIOs support both a centralized and a decentralized team at the same time.

"The bimodal enterprise has two modes of operation," Schlegel said. Mode one is a top-down approach focused on enterprise architecture and security. According to Schlegel, this is the world of performance management where standards matter and everyone is working toward the same common goal. Mode two is an iterative, agile approach focused on analytics and experimentation. Schlegel called this "the domain of exploration," where big data and advanced analytics live, and where failure is not only tolerated, but encouraged.

While CIOs might assume a bimodal IT strategy to be cost prohibitive, Schlegel said the opposite is true. "What we call 'fluid' in mode two is called 'an exception' in mode one," he said. "By eliminating those exceptions and pushing them to self-service, we estimate that almost half of production reports can be moved from an environment where they're produced all of the time and rarely used, to a self-serviced environment where they're rarely used at all and hardly needed."

Hadoop in two years?

During the summit's customer panel, Schlegel asked the audience if Hadoop, a distributed computing framework, will play an intricate part of their analytics infrastructure in two years. The poll response of the 400 or so participants was split pretty evenly: 46% agreed with the idea; 41% disagreed; 13% remained neutral.

"I find this surprising," said Debashis Saha, vice president of global data infrastructure at eBay Inc. He encouraged attendees who disagreed to revisit Hadoop, which he described as a very different technology than it was two years ago. The key to Hadoop is its distributed computing capabilities, he said, which can be a powerful engine for large-scale analytics.

Tara Paider, associate vice president of IT architecture at Nationwide Insurance, agreed, saying Hadoop will play a vital role in the company's analytics program going forward. Nationwide is using the technology to offload workloads and reduce cost -- a primary Hadoop use case for businesses. But its Hadoop cluster is also currently supporting 17 active projects "all around exploration, research and analytics," she said.

Say what?!?

"Getting value out of IoT is going to require the very best real-time analytics chops you have." -- Chris Howard, analyst, Gartner

"Data is valuable, but the story you tell and the action you take with it is what pays the bills. -- Geoffrey Waldmiller, vice president, ecommerce, MGM Resorts International

"Lack of communication means duplication." -- Michelle Lacy, visual analytics and integration lead, Monsanto Company

"'Big data' is overused, and it's killing the meaning. But we will still be talking about social media, voice data, content, sensor data in two years. We better be. It's virtually untapped." -- Tara Paider, associate vice president, IT architecture, Nationwide Insurance

"You do not bring data to a knife fight." -- Tina Nunno, analyst, Gartner

"Leadership is being a collaborator with a direction." -- Francis Ford Coppola, filmmaker, wine maker, hotelier

Welcome to The Data Mill, a weekly column devoted to all things data. Heard something newsy (or gossipy)? Email me or find me on Twitter at @TT_Nicole.

Next Steps

Previously on The Data Mill

Is Apache Spark heir to the MapReduce throne?

Computer augmentation is better for business

Data lake cries out for security


Dig Deeper on Bimodal IT organization