Want to sum up data warehouses in one word? Try "hard," based on results of a recent survey that found that 68 percent of tech execs scaled back plans or requested additional funding because of time and budget overruns in DW implementations -- this despite the fact that 62 percent of them had factored overruns into their planning.
Two-thirds of respondents acknowledged encountering "unanticipated" problems during DW deployments. Thirty-six percent of them described DW projects as "incredibly hard." (OK, so we didn't go far enough in our opening paragraph.) A recent DMReview article listed dozens of "gotchas," grouping them under four main categories: money, design, heritage and data.
Such problems are not too surprising, considering the long and winding road followed by many companies during DW implementations. As described in a B Eye Network article, the process often looks like this: data mart, application data warehouse, functional data warehouse, Phase 1 enterprise data warehouse, and finally, Phase N enterprise data warehouse.
While big-bang approaches to DWs are generally doomed to failure, according to the article, most companies can skip the data mart stage.
Increasing demands on DWs also lead to more complexity and cost, according to this Intelligent Enterprise article. Among the demands: increasing volumes of data and queries, more users and a desire for reduced latency times.
With all of these obstacles, why even bother with DWs? Maybe it's because the payoffs can be so great for those who nail it, as evidenced by companies like Harrah's and Lowe's.
The data warehouse at Lowe's, combined with business intelligence tools, has helped the retailer solve problems as diverse as where to locate cash registers in stores and which product returns are likely fraudulent. Data analysis also helped Lowe's streamline the process of collecting delivery charges from stores, resulting in a $30 milion boost to its bottom line.