Data Integration Remains a Major IT Headache
Study shows that data integration is still costly and requires a lot of manual coding.
In some ways, managing data is as much a political question as a technological one. Given that it's Labor Day, I thought it'd be worth considering the politics of data integration.
Personally, I come at this question as a journalist and writer. I absolutely believe in freedom of the press, and more specifically, the type of open information we get from sunshine laws. I think the fact newspapers are dying is a bad thing, because it's hard work to uncover what politicians want hidden, and I'm not sure bloggers - particularly unpaid bloggers - are up to the task.
Enter the Internet, with its rallying call: "Information wants to be free." Theoretically, I absolutely agree. But then again, I also want to be paid for finding and distilling that information.
I think the same holds true for corporate data. Sure, companies want it to be a corporate asset - they want data to be free - at least within their own borders. We all understand that, in theory: Everybody can do more if everyone shares data, which after all, really belongs to the organization. And aren't we all on the same team? Rah-rah-rah, sis-boom-bah!
But the first time you don't survive a layoff or someone steals your ideas or your source won't return your phone call because too many people from your organization keep calling, well, then you start to suspect that maybe your mother was wrong and sharing isn't always the right thing to do.
Data may want to be free, but it turns out, everybody likes to be paid - whether that translates into political capital, control, real dollars or just not having bridges burned by someone in another department. What's more, if everything I know is shared with everyone else and institutionalized by placing it within the corporate IT systems, well, I've just given up a big part of what makes me valuable.
Sure, it's good to be a team player. But most of us are also a little capitalist since the time we're two and learn the power of the phrase, "It's mine!" We all love to deride politics, but when you come down to it, politics are human nature and part of every organization. And, when it's your job on the line, yes, politics are personal.
Now, in some organizations, labor rallies around the collective. That's why we have Labor Day-to "exhibit to the public 'the strength and esprit de corps of the trade and labor organizations.'" But in today's information and services economy, that ideal of the power of unions has given way to the power of individuals, which raises the question: If you take the ideas of capitalism down to the micro-level in today's organizations, can we really expect to socialize data in our ber-capitalist society? Does data turn even labor's rank and file into capitalists?
Jim Harris discussed how socialism and capitalism play out in the data management world on his OCDQ (Obsessive-Compulsive Data Quality) blog:
In enterprise data management, one of the most debated ideologies is whether or not data should be viewed as a corporate asset, especially by the for-profit corporations of capitalism, which is (even before the Cold War began), and will likely forever remain, the world's dominant economic model. My earlier remark that data is an asset only if it is a shared asset, across the silos, across the corporate culture, is indicative of the bounded socialist view of enterprise data. In other words, almost no one in the enterprise data management space is suggesting that data should be shared beyond the boundary of the organization. In this sense, advocates, including myself, of data governance are advocating socializing data within the enterprise so that data can be better capitalized as a true corporate asset.
He explains how Google and Facebook manage to make money off of free information, including data by the likes of you and me and anyone else who uses their "free" services. And in an unexpected turn, he links this to master data management, explaining how it all relates to their redefinition of customer as a "product." That part was fascinating, but it gave me a major headache and, frankly, left a bad taste in my mouth.
OK, fine, but that's Google and Facebook, right? Well, actually, even defining data can become very political if you're not careful. For instance, David Loshin wrote recently about the many meanings of the term "customer" and the tension between master data's mission to create a "single view" of a customer when there are so many different and valid definitions of the word within any given organization. If you're not careful, how you define master data could, in effect, disenfranchise some divisions.
"... let's think about this a little more carefully - if there are different meanings of customer, a single view introduces a risk of boiling out the (probably *relevant*) facets that are important to each consumer of that data," Loshin writes.
You can resolve this conflict with technology by using multiple master data sets that serve different groups. But then again, you've got silos. I look forward to his next post on the topic to see how he recommends solving the conundrum.
But it's worth remembering that there is a very real political challenge here, and it's not just petty politics. On one side, data represents sales bonuses, advancements and possibly even people's job security. But on the other, data represents growth and the future of the organization.
Harris theorizes that it's the Cold War in small. He may be right. If so, my question is: Can IT and data integration bring down that wall?