Re: Top Ten Errors In Data Wareousing

In article <3i8duh$86b@newsbf02.news.aol.com>, pineryjim@aol.com (PINERYJIM) says: >Now that we apparently have an institute dedicated to the practice of data >warehousing I'm wondering if the term also has an institutionalized >definition. Just what is it you do when you warehouse data? Or, from a >different perspective, what kind of data does a warehouse contain? Or, am >I missing the point altogether? I claim no particular expertise, but in my opinion a data warehouse is a *new* thing. Regular databases are optimized for transaction processing which implies a high proportion of "create or update" activity and, we would hope, not a lot of full table scans and long- running queries. A data warehouse is a repository for static data. The tables may be updated monthly, weekly or daily, but not on-line. Warehouse tables typically have lots of indexes and may be tuned to accommodate the most frequent sorts of queries. Some folks even recommend (horrors!) de-normalizing tables to the extent of, at least, pre-joining lookup tables to create faster access at the (to end-users) negligible cost of redundant data(!) in the warehouse. The really *new* thing is that warehoused data is intended to be accessed by end-users so they can get the data themselves and use desktop tools to turn the data into the information they need. End-user empowerment -- what a concept! This thing could do away with the typical MI application backlog (did I see 7 years cited somewhere) if implemented properly. Bill Parmenter Sr Software Engr Purdue University