John Zachman, the father of information architecture (and a really great guy to boot!), reported at a recent conference that 45 percent of all computer reports produced are never used. Of the remaining 55 percent, 70 percent are used for six months or less. He also indicated that 70 percent of computer printouts are used to re-enter data into other systems. One of the implications of these facts for data warehousing and business intelligence tools is that reports are not necessarily what they used to be. Business people want the flexibility of manipulating information directly. No one has time to "turn off" reports they are no longer using. In effect, the business is changing, and we have to help cause a culture shift in our organizations regarding the usage of information. When building a data warehouse, its implementers have a decision to make regarding user access--to control it or not to control it. There are reasons to do each, and making a choice about the direction to take regarding control of the end-user environment is one of the most important choices to be made in any warehouse implementation. It comes down to an issue of control versus empowerment.
Please realize that both alternatives have merit. Pursuing one over the other depends in large part on the overall information culture of the organization, including which information behaviors are rewarded or incented and which are not. For example, is fact-based decision making rewarded? Is it accepted to base decisions on an appropriate analytical process or rather on intuition? Is de-centralized intelligence valued or are all decisions made centrally? Answers to these questions will help develop the appropriate course to take regarding the deployment of business intelligence tools within an organization.
Controlling the user's process of querying the warehouse involves providing structured access--predefining queries anticipated to be utilized most often and providing those queries/reports in a library, or developing access "applications" where SQL statements are developed in a program and only certain queries are presented to the user. The idea is to tune the database to provide acceptable performance for these queries, based on user expectations. Control works very well when a majority of user queries can be anticipated.
Empowering users involves allowing fairly unrestricted access, by allowing them to perform joins based on foreign keys and by allowing (but shielding the user from) the creation of individual, specialized SQL statements through BI tools. Empowerment is fundamentally different from the control model in that response-time performance cannot be guaranteed, and users have to be counseled that performance can vary based on what they are trying to do. Empowerment works well when a majority of user queries are ad hoc and cannot be anticipated.
Reasons some organizations control a user's data warehouse access include more efficient CPU usage and the need to provide fairly consistent response time. In fact, repetitive queries can be tuned to achieve very good response time, which pleases the user community, the database administrators and the capacity planners. Some variable constraints can be provided to customize the query results for a given geography, time period, product line, etc., while avoiding a "free for all." But the problem with the control model relates to John Zachman's statement. The business is changing, and business people need to see changes as well.
The major reason other organizations promote user empowerment is to provide business flexibility. With empowerment, users can pretty much do whatever they dream up. However, as mentioned before, they need to expect wide variations in response time. And they need to take on responsibility to know their data, determine how to make a "query" out of a business question and to determine the query's "fit" for the business problem at hand. Empowerment is about information democracy, and it's about trust--trust that the data will be used correctly, trust that the query won't dim the lights in the building while trying to complete, trust that the users know what they're doing. And how is that trust developed? By training users about their data and about query techniques. They need to know the best way to perform queries they are interested in and the consequences of inefficient query development (such as cartesian products). They need to try some things, make some mistakes, get some help on how to improve query performance and try again. Not everyone will be "worthy" at first, but they will at least be able to claim self-sufficiency. And with help they will get better over time.
So when determining the "oh-so-important" strategy on how users will get access to their data, think about what users need to do, what information behaviors are incented within the culture and what users are capable of doing. And if you choose empowerment, provide training and a help-desk function to assist them in their independence. After all, it's a matter of trust and education and guidance and self-sufficiency.
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access