David Linthicum has a plan to fix the federal government's data-integration problems. After all, data integration is easy, he writes in a recent Informatica blog post, so the key is planning and selecting the right technology.
It's not that Linthicum thinks the government doesn't have some level of integration; clearly it does. But it doesn't seem to be working, he suggests, pointing to the recent failure of systems that allowed suspected terrorist Umar Farouk Abdulmutallab to board a plane and launch an attack Christmas day.
As I shared earlier this week, a Federal News Radio reporter says an unnamed former senior intelligence official blamed inadequate data integration between agencies as the reason intelligence failed to "connect the dots" on Abdulmutallab. You can find more details about the disconnect between agencies and information in a Eugene Robinson column published today on The Washington Post's site.
Lithicum picked up on the same Federal News Radio report and agreed with the unnamed senior official's assessment:
After looking at the facts of the case, I have to agree. Indeed, September 11th and other attacks could be avoided if there was a common understanding of information between agencies. We seem to be missing patterns among the larger amounts of information collected, patterns not understood because we lack a well understood data sharing and data integration policy and practice between those who own the relevant data.
It's concerning, because as any data-integration expert will no doubt tell you, this is a problem that can be solved-at least from a technology perspective. But, as always, the real difficulties may be more about internal politics, fiefdoms and management more than technology.
If you're in the private sector, try not to be smug. As Linthicum points out:
We also see this in corporate America, but the end results are much less tragic.
He suggests an eight-step plan for creating a data-integration plan that will work not just for the government, but for businesses. It starts with step zero, probably so-named because it's a fundamental step, but it's outside the domain of the actual data integration: Have a single entity control the IT budgets until this problem is addressed.
The problem with this suggestion is the involved organizations will probably balk there, before they've even progressed to step one.
But, assuming your company or the government can make it that far, it's a simple plan. Effectively, he's prioritized the well-known steps you'd need to take to solve data integration. That's more useful than it sounds, because one problem organizations tend to run into is that they try to solve everything at once, rather than taking the time to ensure the smaller steps work, as Francis Carden, founder and chief evangelist of OpenSpan, recently wrote in response to my post:
Technologies exist today to enable a crawl walk run integration strategy. The problem often with technology is that everyone seeks out the perfect solution (and skip years of immediate benefit) instead of applying instant-on solutions that can cover your back-side and provide a progressive step to perfection.
It's useful to know the order in which these things should be addressed. Check out Linthicum's suggestions.
And on a personal note to the integration community, wouldn't it be great if you could come together to help solve this problem? It could help save more than money and time... it could help save lives.