Big Blue's database dilemma

Mapping a conventional business onto a U2 database is straightforward and reflects the Post-Relational (PRDMBS) concepts put forward by Ted Codd in the mid-nineties. Oracle on the other hand is based on Codd's late-seventies model. The trouble with juggernauts like these is that they are so heavily laden with layer upon layer of superfluous gloss (integrated modeling, integrated 4GL,...integrated everything) that it causes the transportation level to lie "too deep in the water" to be maneuverable.

Consequently when the environment changes and they pass through a flotilla of highly maneuverable models (yes like U2) they cannot compete or out-maneuver them so they give a loud blast on the marketing horn and plough straight through. The trouble is one day they will fail to recognize that the blip on the radar screen is not the usual toothless vessel and a Titanic error will be committed.

The thing is, you have to understand the products to be able to understand the benefits. As our businesses change, extending a data set by adding a handful of new attributes and creating a few new files and indices takes minutes and does not warrant software rebuilding at all in most cases where in a juggernaut environment this is a major upheaval. Of course there are always quick and dirty expedient measures that can be taken, but this is a buy-now-pay-later approach, which tends to introduce major inefficiencies and forces regular re-engineering and database tuning exercises.

U2 is more of a hovercraft by comparison--highly maneuverable, portable over hugely disparate terrain and about as "open" as PRDBMSs get. There used to be many vendors of this product range but Raining Data and IBM is the only major players left.

IBM now has a dilemma. Whether it wanted U2 or not it's got it. If it capitulates and declares a "shut-down" the users will migrate to Raining Data's D3. If it decides to exploit the technology, it will need to throw some much needed R&D into the arena to compete with Raining Data's latest Objects Development products.

If IBM decides to exploit the user base, it will shoot itself in the foot and this will be taken as capitulation--this will be even worse than Informix's approach because they at least moved U2 forward to some extent. Well, they had to, because U2 is the underlying technology for Datastage. This last fact seems to have eluded Russom.

Then there's IBM's existing patronage of U2 technology--a recent revelation from Janet Perna, of IBM's Data Management Div., acknowledges that one of their Web sites (www.printers.ibm.com) is a U2/Redback application (there could be a touch of "Victor Kyam" here--saw it, liked it, so I bought the company!).

Much of what Russom says is plausible. Ascential did indeed "cream-off" the best products and left the pile of databases for someone else to sort out but most speculators fail to realize that Ascential took with it a copy of the U2 Blueprint which underpins Datastage. So if we're playing "spot the dinosaur" maybe I can remind the author that anyone can save time on research by employing liberal quantities of prejudice, as he has, to expert effect. Unlike him I do not have the advantage of a PhD in Music, but maybe my experience of corporate IT systems will lend a little weight to my views and help redress the balance.

There has been a great deal of drivel published on the IBM/Informix deal lately from "experts" presenting their predictions as fact. Many of whom have been predicting the demise of U2 for decades, I might add. One day they will surely be right, but chisel those predictions in stone chaps, or the archeologists may not acknowledge you when the time comes.

The fact is U2 is one of the best values for money, scalable Unix/NT solutions you can find in the 10-5,000 user arena, and offers its customers the ability to change their minds on functionality with minimal effect on the maintenance budget. Maybe this explains why it's so strong in the SME market place. For organizations in a state of flux, expanding rapidly, and exploiting new markets, flexibility and versatility in corporate strategy and structure must be mirrored in the database supporting that organization. The "leading" DBMSs can prove so expensive to modify that the PRDBMS (aka Multi-Valued ) model is very competitive indeed. Unnormalized (FNF), nested data is what U2 thrives on, not that one should skimp in these areas--but U2 has such a forgiving nature.

There are other products involved in the IBM deal, so I hear, but there are other authors better qualified to comment on those than I, and who are similarly on the payroll of no interested participator or competitor.

Ray Jones is a former chairman of Ardent User Group in the U.K., and currently is an excutive with PHS Group.