Got a thought or idea for our BI Bloggers ? Email biinactionblog@ebizq.net
BI in Action Blog
|
« Teradata's New Course Up the Stack | Main | Survey Finds Open Source Wide, Not Deep, at Data Sites » October 19, 2007Real-Time Data Availability? Why Not 'Good Enough' Availability?
Listen to the vendors, and you can be forgiven for thinking that just about every business now runs on a real-time, analytical, automated decisioning, neural networked infrastructure. Time for a reality check, and this one was provided by Donald Feinberg, vice president and analyst with Gartner. At last week's Teradata Partners Conference in Las Vegas, I had the opportunity to hear Feinberg talk about what's important, what should not be as important, and what isn't emphasized enough in BI these days. One theme Feinberg emphasized -- reflecting a theme I heard throughout the conference -- was that data needs to be closer to the infrastructure that supports it. A trend Feinberg is seeing is that enterprises are "pushing DBMS code closer to storage, and sometimes into the storage." This is a positive trend, he says, because "the more DBMS code is moved closer to the storage, the less data movement that is required." "You're no longer pushing a basketball through a straw," he added. Another issue that many companies are coming up against is the need to hang on to data -- sometimes a lot longer than necessary, Feinberg says. "You don't need to keep all your data forever," he advises. "Why would you need market basket detail for 10 years? Are you going to send a letter to a customer that bought a Bic pen 10 years ago, and ask them why they haven't bough a new one?" This is a good point, especially in the current environment where everyone is nervous about compliance. I even ran into one company where the IT administrator admitted that their policy is now to hang on to every email they've ever generated, and intend to keep it stored somewhere forever. Another point Feinberg raised is the sense of urgency many businesses feel about making "real time" data available whenever and to whomever wants it. This is an expensive proposition that will deliver little or no added value to the business, he points out. "You don't need instantaneous loading of things unless you're the New York Stock Exchange, and you're looking for fraud in stock trading." "Good enough" data availability may be just enough the fit the bill for most companies, he says. For example, "the time for phone service activation has gone from two weeks to six hours -- and that's 'good enough,'" he points out. Feinberg also talked about the benefits of establishing a business intelligence "center of excellence" within the company to move projects forward. While some may see a center of excellence as another bureaucracy, it has just the opposite effect, he believes. A business intelligence center of excellence "takes the politics out of business intelligence altogether," he points out. "If you put in BI systems, you have to decide which department gets the first one? Every department will get mad because they wanted the system first." A center of excellence can look at the business drivers for BI projects and make the call as to which departments will see the first deployments. Posted by joemckendrick in Management | Digg This | Add to del.icio.us Trackback Pings TrackBack URL for this entry: |
