My colleague and friend Jim Hodel has published a great piece on data abstraction layers and data quality in the June 2006 issue of DM Review. Jim and I met at a common client and have collaborated on elements of what he describes in his article for that client's data architecture and administration initiatives.
I've long maintained that "apps are crap without good/valid data," and Jim outlines a great layered straw-man approach to get organizations started on organizing data properly and improving data quality in the process. Well worth a look for those with simmering data administration/abstraction issues.
i found your blog only today. this post alone made the trip worthwhile.
while i am an architect as opposed to a DA, i found the ideas in the article outstanding. simple, yet profound: realize the lower level as singleton component units containing data that should be promulgated as an enterprise vocabulary, rather than copied in various systems across silos. then decorate them, and combine them as you move up.
Posted by: dragongrrl | July 07, 2006 at 11:17 AM