Within the first two weeks of any undergraduate algorithms class, students learn about the Traveling Salesman Problem. I’ll let the web fill you in on the gory details of the problem, kindly refer you to Dykstra for one tidy solution and cut straight to the lesson: often the solution that offers the initial lowest cost turns out to be the most expensive solution in the end. The problem is designed to teach neophyte computer scientists the lesson of planning ahead, typically all the way to the final end state, before deciding on a solution. The solution is applicable to everything from database query optimization, to optimal shoe-lacing (if you’re into that) and yes, even to the optimum data integration architectures.
Enterprises are asking a lot of their data and their IT organizations these days. Smart organizations understand that harnessing data is more than just generating stale reports that never get read. Timely, trustable data, available when and where it’s needed, in the right form for the job at hand can drive top-line revenue growth, improve operating margins, facilitate M&A integration and satisfy auditors performing compliance checks. However, few companies possess the foresight to architect their systems to deliver all of these benefits. In my conversations with customers I see everything from the data equivalents of newborns (hand coding solutions to the most immediate task at hand) to the wise grey hairs (carefully plotting out service level agreements, latency requirements, data growth assumptions and other factors before making a move). The primary difference between the two is foresight. The newborns lack the experience to know that there will come a time when almost every implicit or explicit assumption they are making about their data, its use and the operating environment will all change. In short, they make the same mistakes beginning computer scientists make; they wrongly assume lower start costs equals lower long term costs.
In future blog entries I’ll discuss best practices in enterprise data management and how they drive real business benefits.
