Last week I attended a Data warehousing course taught by the Kimball Group. It was a great course. Going into it I had only the most superficial of understandings of data warehousing concepts, specifically regarding dimensional modeling, but now feel that I have a pretty workable understanding. I highly recommend the course to anyone interested in the topic, not that Ralph Kimball needs a recommendation from me.
In addition to learning about dimensional modeling, I made another tangential observation. It felt like data warehousing/business intelligence is largely figured out. In that course we must’ve reviewed dozens of case studies ranging from straightforward to complex across a range of industries, but all the solutions seemed to break down in to a few well established ‘types’ and patterns, patterns that have existed (evolving) for decades. My observation may be a bit starry-eyed and I’m not at all suggesting that implementing a DW/BI system isn’t complex and challenging. Nevertheless it appealed to me, as a developer, because I’ve never gotten that impression from a software training, conference, or seminar. On the contrary, I find more often than not we’re being introduced to new approaches and patterns suggesting that we should abandon our old way of doing things. Instead of breaking down our application architectures into common types, we’re constantly creating new types and patterns, a pattern/type explosion.
Maybe we need an application architecture toolkit?