Praise Be To Codd

October 1 1993 by Robert Bittlestone


Kevin: Bob, cancel all your appointments. I simply have to tell you about a wondrous event that has just changed the IT world forever.

Bob: One of your computer projects came in on time and within budget?

Kevin: No, not quite as radical as that. Ted Codd’s just published his new “Twelve Commandments”! Codd’s the ultimate IS guru. In 1985 he redefined the world of transaction processing by writing his 12 laws of relational data base management. Now he’s done the same for online analytical processing. Twelve more commandments, every one a winner!

Harry: You’ve lost me, Kevin. What’s a relational data base or an analytical process?

Kevin: A relational data base is how we store transactions. One of our salesmen in Oregon sells a product to a customer in Portland, and we have to bill him, so we need to store the data. We could write it out in a line-salesman name, customer name, product name-but that would be wasteful: We’d end up repeating the names of our salesmen and products thousands of times across the file. A relational data base lets you keep different files for these different things and link them together in ways that avoid data duplication.

We all thought Codd had cracked it with his first commandments. If you got your relational data base right, you could answer any question, including, “Tell me the number of left-handed salesmen in Minneapolis who speak Greek.” It was awesome. But after a while, we began to have doubts. People didn’t ask the questions we thought they would. Instead, they asked about trends over time, comparative totals by business or product, or variations against different planning scenarios. We tried to get that into our computers, but it killed the hardware-instant lockup, gigasecond response time, thrashing partitions, negative CPU utilization, evaporated water-cooling, piles of smoking iron all over the floor, that sort of thing. And here’s the worst part: Doubling the CPU power made no difference.

Harry: This didn’t happen first in 1988, by any chance?

Kevin: Spot on, Harry! How on Earth did you know?

Harry: That was the year you did a board presentation explaining why we had to spend another $15 million upgrading the data processing center.

Bob: Are you telling us that we spent all that money for no good purpose?

Kevin: Bob, hindsight is a wonderful thing. We learned a lot from the experience, and we used the spare CPUs for development, but the extra horsepower didn’t impact the analytical problem.

The new commandments will. They’re called the rules of Online Analytical Processing or OLAP. The OLAP rules are different from the relational data base ones. There’s one about multidimensionality, for example, that says you may want to look across products or countries or time periods or reporting lines without having to make that choice in advance. There’s another about transparency that says you should be able to access analytical data in the same way regardless of where it comes from or which application you’re using. There’s a third about accessibility that says the OLAP tool should be able to aggregate and convert the data as it goes along. There are others about efficiency and not wasting disk space. Overall, it’s a real shot in the arm for IS professionals; a chance to break into a whole new area of executive support activity.

Harry: Kevin, you probably know that the financial systems people who report to my controller don’t talk to the IS people in your group. Does this mean that at last they’ll have something in common?

Kevin: Absolutely. We never understood what you people did with your consolidation systems and your budget models. Now we’ve found some paths through the jungle, and we can start to help you.

Olaf: Kevin, whenever you talk about “helping” us, I usually reach for my capital expenditure veto, but on this occasion I’m interested. What will it do for my strategic planners?

Kevin: For a start, we’ll be able to support planning scenarios that really link into the operational side of the business.

Olaf: Let’s not get too radical. I can’t believe anyone would want to do that.

Bob: What should we do today that’s different from yesterday?

Kevin: We should start by making sure. all our finance and IS people know about these developments. You can get Codd’s original paper from his company, Codd and Date in San Jose. There are also papers available from the Metapraxis people in New York and London that discuss these issues from a managerial perspective. I guess once we’ve assimilated the thinking, we should bake these new directions into our IT policy.

Harry: And save the next $15 million?

Kevin: Absolutely! This new thinking spells death to the mainframe as an analytical engine.           

Bob: Analytical engine? Haven’t I heard that before? Charles Babbage, the Victorian engineer . . . The kids dragged us to the London Science Museum on our August vacation in the U.K., and there was a whole computing contraption with gears and everything. Is that what you were planning to buy for us next? I liked the look of that one. A sort of IS version of the Patek Philippe watch. A touch of class beneath the glass. Ah, well. Plus ca change…

Robert Bittlestone is founder and chief executive of Metapraxis, a London and New York-based consulting group specializing in performance measurement and strategic control.