Eye on TDWI – August 2003

The day before The Data Warehousing Institute’s World Conference kicked off in Boston, I attended the wedding of two good friends. The chance of their meeting had been a long shot. He was a
successful, but shy architect. She was a physician with not that much free time. Enter www.match.com. Talk about looking for information in all
the right places! I thought to myself, what a great way to kick off the week with a bunch of data hounds. I toasted the new couple and headed for the airport for the red-eye to Bean Town. I arrived
just in time to catch the first session.


I have been to a lot of TDWI conferences, but I had never taken a management track. However, I had been recommending these sessions to some of my clients, and thought that I should sit through a
few of them to make sure we were all on the same wave-length.

Maureen Clarry and Lorna Rickard’s workshop, “Power, Politics, & Partnership in Data Warehousing,” was first on my dance card. Mercifully, Clarry and Rickard herded us through a lot of large
motor activity throughout the day, which, after the red-eye, managed to keep me from lapsing into a coma.

Clarry and Rickard focused on the art of overcoming the negative impact of politics that can often sideline a data warehousing program. They broke us up into groups of customers, execs, middle
managers, and grunts – and then gave us some basic roles to play out within the time constraints of 12 minute days, i.e. no coffee, no lunch breaks. We weren’t told to be nasty to each other, but
the changing demands of our customers and our organizational responses induced enough stress and frustration to simulate the real thing. This reality show for data warehousing teams pointed out how
easy it is for each group to make up stories about the others, take things personally, get mad, get even, and/or withdraw. Each negative reaction endangered any sense of partnership that the groups
needed to provide goods or services or to get what they wanted. Clarry and Rickard held that the key to success was for each group to actively engage the others in our mutual projects. I think that
each group gained an appreciation for the stresses placed on the others, and we all got a tweak in our emotional IQs.

By the end of the day, my addled left-coast brain was suffering from a significant lack of sleep. I turned in early. Ralph Kimball was giving the keynote that next morning, and I wanted to be up
for it.


Kimball’s keynote was titled “Total Cost of Ownership Starts with the End User.” However, he quickly blew past a discussion of explicit data warehousing costs – hardware, software, labor, etc. –
to tackle the larger, more significant costs of failing to deliver the right data to the business at the right time. He suggested regularly canvassing clients for new requirements and trolling for
new sources of business metrics. As Clarry and Rickard had done the day before, Kimball stressed the need for partnership between IT and its clients. As he has done in his articles and books,
Kimball warned against non-integrated data and non-verbose data (reliance on codes, etc.) that limit the use of the data to a small customer base. He also warned against locking data up behind
tools that unnecessarily limit user access and analysis. Kimball exhorted us to think like publishers, remain flexible, and live amongst our customers.

Those Kimball followers in the keynote audience that were looking for some sort of neat costing model to go along with their dimensional models may have been disappointed. I think that dimensional
modeling fans often get to be overly fascinated by the structural challenges of designing the data warehouse and brush aside the significant soft issues. However, we ignore those soft issues at our
peril. Kimball, the thought leader in dimensional modeling, was the right person to deliver that message.

After the keynote, I dropped in on Jill Dyche’s “What Business Managers Need to Know about Data Warehousing.” Actually, it could have been titled “What We All Need to Know …, ” but that
would have scared off the managers. Dyche’ brought her considerable experience to bear quickly and recited a list of lessons learned early on in the data warehouse game. I was impressed with the
way she had crafted her session in business terms right from the get-go. Dyche’ then challenged some myths. She asserted that “If you built it, they probably won’t even know,” referring to IT
only DW projects. I liked Dyche’s focus on data warehousing applications, which included “end-user functionality, presentation, and data.” All too often we data dogs break down our DW increments
into data centric projects and put off dealing with presentation and end-user functionality for some other time – big mistake. Dyche’ also led us through a quick benefits calculation or two. She
left us with several admonitions: look for sound business value (show them the money), don’t call something the data warehouse that is not a data warehouse (adopt best practices), and promote the
idea that data should be managed as a corporate asset (but you already knew that, didn’t you?).

That evening, night school offered up a gem. Howard Spielman’s “Business Intelligence – Past, Present, and Future – What 200 Years of Executive Decision Making means Today” was a fascinating, 90
minute tour through the history of data visualization. Way back when we didn’t have Excel, the main customers for such things were kings, etc. Now, my idea of a king, is a guy who is surrounded by
really elegant stuff. Spielman’s tour showed us some of that really elegant stuff in the form of charts and graphs. He then took us up through early data visualization practices intended to
communicate ideas to everyday folks, i.e. graphs and 3d charts produced in the industrial age to explain such things as power consumption patterns over time. (I wonder what that would have looked
like a week or so back in New York?). Next, Spielman brought us up to the computer age and showed us ways that tricky (or dumb) analysts have packaged data within graphics to, um, tell a story?

Spielman concluded with an unsettling thought. We have come a long way in terms of building machines to house big data and in designing systems to collect those data. However, in spite of having
more data than ever at our fingertips, we have yet to come anywhere near the elegance of the past in making the connection between those machines and our brains. Computer assisted data
visualization is still in a primitive state and Spielman claims that the industry’s big challenge today is improving the interface between screen and brain.

Spielman was a fascinating person with a lot of obvious experience. I invited him to have a drink (diet coke for him, single malt for me) and we ended up talking well into the night. That’s one of
the things about TDWI conferences. You meet new colleagues and in a flash, you can find your horizons stretched well beyond expectations.


After a quick breakfast, I sat in on a TDWI forum on business/IT partnerships in Business Intelligence – sort of a conference within a conference. One of the speakers was Louis Barton, an executive
vice president at Cullen/Frost Bankers Inc., who related how his firm had successfully organized into three groups to support its business intelligence environment. Those groups had well defined
responsibilities for data privacy, data warehouse, and data stewardship, and they all shared responsibility for the BI environment. Interestingly, the folks in data stewardship were responsible for
data modeling and meta data plus addressing and monitoring data quality issues. Hmmm. All of you refugees from data admin groups that have suffered budget cuts and down-sizings, take note.

That afternoon, I took some time to visit the vendor show. The folks in the Cognos booth were previewing ReportNet, which Cognos will announce early next month. This represents a significant
upgrade in the Cognos BI tool suite. Knightsbridge, a consultancy specializing in large scale extract, transform, and load (ETL) applications had a magic guy that was very good. It fit with the
things that Knightsbridge does with big data. Speaking of ETL, several years ago some analyst proclaimed that ETL technology would be taken over by the database vendors. It was obvious by the
vendor show that that hasn’t happened yet, and may not happen anytime soon. ETL vendors were out in force, offering everything from high end to budget level tools.

By the time evening rolled around, I was ready to start looking for a good jazz club. However, the folks at Brio read my mind. I dropped by their hospitality suite before heading out and found the
sweet sounds of Brazilian jazz coming from the flugelhorn of

Johnny Souza with the Ray Santisi trio.

What a find!


I thought that I’d break out of the management track for day and catch Michael Gonzales’s “Hands-On Data Mining” lab. Although Gonzales appeared quite normal in the conference brochure bio, his
unusual high energy, breadth of knowledge, and reputation for being a truly fun guy ensured that these hands-on sessions would be sold out about the time the ink dried on the conference schedule.
Over the course of the day, Gonzales made the subject quite accessible even for the mathematically challenged. He started out with the basics and then led us through a series of simple exercises
with three different software packages from Microsoft, Teradata, and SAS. Whew! Great workshop!

That afternoon I dropped in on Wayne Eckerson and Cindy Howson’s session on “Evaluating Business Intelligence Toolsets”. They took a sampling of major BI vendors -Cognos, Business Objects,
Crystal, and MicroStrategy – and led us through a pretty rigorous process of evaluating not only the toolset capabilities, but also the vendors themselves. Although some information had become moot
due to the purchase of Crystal by Business Objects, the primary value in the course lay in the framework that they gave us, not the sampling. Programs looking for BI toolsets will still need to
apply their own requirements to that framework or something similar. There is no such thing as a free lunch.

By evening, I was getting hungry. I found a trombone player and we went to the California Pizza Kitchen to discuss the challenges of accommodating changing historical relationships in type II
hierarchies. If you have any ideas on the subject, let me know. If there’re good, I’ll tell you who the trombone player is.


Over breakfast, Barry Devlin delivered the second keynote of the conference, “What’s Next for Data Warehousing?” Devlin has been in the forefront of decision support systems architecture since
the mid ’80s. In his recent reading of the tea leaves, Devlin gave us a vision that included federated decision support data from sources including the data warehouse, operational systems, newly
acquired companies, the supply chain, and the customer. Unfortunately, the term “federation” has caused the shields to go up for some. Devlin was careful to underscore that data federation was
not a substitute for the history rich data warehouse, (in spite of what middle-ware sales reps would have you believe). The foundation for data federation is based on a unifying meta data layer and
the technology to support access to target sources using this meta data store. With the demand for decision support access to real time data on the increase, the data federation strategy seems to
be worth considering.

I often see fledgling data warehouse programs get into hot water because of a lack of an end-game. Designers want to design, programmers want to code, but unless there is a plan for data warehouse
administration, there is not much point in building the thing in the first place. Jonathan Geiger’s “Managing Your Data Warehouse: Ensuring Ongoing Value” addressed this issue head-on. Although
Geiger has been one of the guiding lights in the Corporate Information Factory camp, most of his strategies would apply equally well to a Data Warehouse Bus Architecture shop. His course covered
such topics as service level agreements, change management, benefits tracking (big, BIG item), plus auditing and assessments. Geiger’s course should be required for all prospective program

That night a I caught Ray Santisi and his group again, this time at the Marriott bar. Good jazz and good friends – what a way to spend the last night at the conference.


Wayne Eckerson and several other instructors put a wrap on the week with a Town Meeting titled appropriately “Putting it All Together”. By then the crowd had thinned out. That gave each of us a
chance to have the experts opine on our individual problems. However, I was surprised at the commonality of concerns. Poor facilities for meta data interchange between tools and a dearth of low
cost-of-entry reporting and analysis suites seemed to be at the top of most everybody’s list. The instructors gave the widespread adoption of the common warehouse meta model (CWM) by vendors a
better than even chance of providing better meta data integration, but it might take some time to come about. We kicked around several strategies for dealing with the high cost of BI tools. Second
tier BI vendors may be worth a look for small organizations with a limited budget.

In the next TDWI Conference coming up in November, Cary White and I’ll be hosting a special interest group on data warehousing for higher ed. Catch you all in San Diego.

Share this post

Tim Feetham

Tim Feetham

Tim is an independent consultant who specializes in data warehousing for small to medium sized businesses. He has worked in sectors ranging from travel, health care, finance and software, to higher education. He helped design the Data Resource Management Certificate at the University of Washington and has taught in that program for more than 10 years. Feetham is also a former senior research analyst for TDWI. He continues to contribute to TDWI publications and events.

scroll to top
We use technologies such as cookies to understand how you use our site and to provide a better user experience. This includes personalizing content, using analytics and improving site operations. We may share your information about your use of our site with third parties in accordance with our Privacy Policy. You can change your cookie settings as described here at any time, but parts of our site may not function correctly without them. By continuing to use our site, you agree that we can save cookies on your device, unless you have disabled cookies.
I Accept