Mike Kavis, a blogger on ITToolbox and Chief Architect, sent me an interesting question about enterprise metadata
He says “My company has huge amounts of historical data. We load 500M new rows a day and keep data for 116 weeks. I am trying to move this company away from being report writers to being enablers of information. We have both Microstrategy and SAS yet we still seem to write a lot of canned reports. One of our challenges is we have not created a logical representation of our data across our various data warehouses and data marts. If we could present data in business terms to our users and point them to our BI tools, I think we could be a much more efficient company and produce more accurate analytics. All of this is leading me down the enterprise metadata path.” Now I am not an expert on enterprise metadata (sorry Mike) but I think I could make some useful contributions to a discussion so, with Mike’s permission, I am going to ask him some questions to help focus my answer:
- David Marco’s definition of metadata is “all physical data and knowledge from inside and outside an organization, including information about the physical data, technical and business processes, rules and constraints of the data, and structures of the data used by a corporation” or a more limited one, focused on “data about data”?
Me, I think the broader definition is more interesting in part because it clearly includes the rules that govern how the organization operates. - Is your end-goal to improve the analytics, improve their usage or make it easier for non-technical users to develop their own reports?
I am not a fan of reports nor do I regard them as a useful form of analytics so my inclination would be to focus on the value of analytics and their usage. - Do you have a Customer Data Integration (CDI) or Master Data Management (MDM) program underway?
Mike, over to you.
Comments on this entry are closed.
1) The broader definition because the business rules pertaining to the data is important.
2)The end goal is to provide both end users (analytics and power users) and developers with access to the logical view of our data. I basically want to hide the complexity of the physical representation of the data and expose derived values. That way, whether people are writing code using SAS, Microstrategy, Java, .Net, or if they are using a query tool, they will all get the same answer. And when the business rules of a metric changes, we change it in one place.
3) Not really. We are using a MDM tool for our initial BPM/SOA projects but we are not currently addressing the enterprise.