Redwood Shores, CA
December 15 , 2004

Oracle 11i.10

Differentiations

According to Oracle, 11i.10 is the only product on the market that is truly capable of supporting a single global instance of the corporation's key business information. They can point to at least one corporation that's proved the point: Oracle. When considering 11i.10, one first has to ask whether they've really nailed what is good about this product.

I think not. There are (and will continue to be) good things in the product. But the people in charge of spin aren't bringing these things out.

To run a global business, you have to have a system that is scalable (check), one with a very, very good data model (check), one that can handle all the complexities of managing local data that satisfies local requirements and is available to the global system (check, I think). All real strengths of the system.

One also has to have a system that supports the processes that the company uses, particularly the critical or strategic processes, without imposing too much management burden. (sometimes yes, but often no, especially when you're talking about the entire corporate footprint). And you

Will this happen to Oracle? Probably. The odds that history offers are poor.

But history is not gravity. Number twos don't just collapse because the forces of nature pull them down. There are reasons why they can't compete, reasons that are located in character, in vision, and in culture, as well as economics.

If the new Oracle does fade from the application scene, I think that the major reason will have nothing to do with the size of its current big competitor or its competitor's great technical prowess. It will have to do with Oracle's own failure to lead.

Now I admit that leadership is a pretty intangible thing, and "failure of leadership" is more likely to be a slogan that Democrats fling at Republicans and Republicans fling at Democrats than anything tangible.

But in the applications space leadership--real leadership, business leadership--has always mattered a lot and for good reason. The people who buy transaction systems are looking to the technology companies to show them how to make material improvements in their business operations. They are saying, "You, Mr. Technologist, need to come up with the innovations that make my business run better."

In my view, Oracle's gradual decline in the applications space over the past few years has been due in large part to the fact that only a few companies have been able to look to Oracle for this leadership and actually get it.

Not that the notion of leadership is foreign to Oracle. In the tool space, they have taken the leader's role and been comfortable in it. The two million people who work with databases look to Oracle for standards.

But in the application space, no. Whether you're talking about 10.7 or 11i, CRM or e-marketplaces, the Oracle CPG initiative (1996) or the customer data hub (2003), Oracle's ideas about how to run businesses have been grandiose, thin, and derivative, and their execution on those ideas has been (understandably) weak. There have been not only better products at other companies, but more understanding of businesses, and better ideas.

So, now that Oracle has swallowed one or two or three product lines of companies whose applications leadership has been better (sometimes), but whose financing has been much worse, can they provide some leadership?

The Information Company

If you ask Oracle, of course, the answer is that they already are leading. (What other answer would you expect?)

The current marketing campaign, they would say, is the kind of crystallization that visionary leaders provide. The campaign, in case you missed it, says that Oracle has become "the information company," the company you look to when you want to store information that is coherent, clean, and easily available. In essence, they are trying to open up a new conceptual space, the information store, which lies above the desktop (Microsoft's) and below business operations (the Oracle customer's).

"The information company" of course provides a complete product line: applications, application server, communications, and database. As far as it goes, this notion of a single stack has some surface appeal (more on this later). But in itself, it's not exactly original. (anybody hear of Netweaver?). And the way Oracle handles it, it is not much of a business idea.

If you heard Charles Phillips' talk at the Oracle user conference, you got probably the single best characterization of the information company. At corporations, Phillips says, too much information is locked away in too many places; not only that, too much of this information is simply bad: inconsistent, wrong, out of date. What is needed is a way of solving this problem, which Oracle provides.

Part of leadership, of course, is understanding the problem. And here, Phillips does pretty well. Phillips himself is a well-educated man, who can cite the legal doctrine of res ipsa loquitur and mention Martin Luther in a speech about technology and get away with it.

And when he talks about bad and incomplete corporate data, he is indeed identifying a problem that has gotten too little attention from application companies. Phillips has talked to hundreds of CIOs and CFOs about their issues. Picking up on this problem and putting it front and center shows that he's a good listener.

But you have to go beyond simply identifying a problem if you're really going to lead. You need compelling, original, and insightful solutions. And here Phillips (and, by extension, Larry) falls short.

The solution is for companies to store their (transaction) data in as central a place as possible. Ideally, this is a single global instance of Oracle applications; failing that, you should use a data hub that provides a data model for storage of strategic information (on an Oracle database).

Every time I hear this, I have the same reaction: huh? This isn't an original, compelling, insightful conclusion. This is the same idea that SAP came to the market with fifteen years ago.

Frankly, even fifteen years ago, it seemed like a pretty vague, hand-waving sort of idea. As the many SAP customers who tried it found to their pain, simply having a central data store does nothing about data quality; it will accept garbage just as easily as it accepts good stuff. And centralizing a transaction system takes a huge amount of effort, because one has to reconcile operational processes in all parts of the globe. Especially for companies that have no reason to centralize operations, it's just not worth the effort.

In those intervening fifteen years, we've learned a lot about the problem. The people who have learned the most, people like Shell, IBM, or duPont spent a lot of money on that education and are still not finished.

If you're going to lead in this area, at the very least you have to show that you understand what they understand, have learned what they've learned, and are now providing innovation relative to that. You need a diagnosis of the problems that emerged over the past 15 years, and you need a cure that takes account of what has been learned.

And this you simply don't see. Instead of showing that he understands the problems facing businesses that have tried this, Larry blames SAP's inadequacies for the problems and points to his own global instance (five years plus in the making) as a just example of why it can and should work.

Make no mistake; he is right as far as he goes. Scratch any Oracle executive, and they'll tell you that their single global instance makes a difference. At an analyst session, I asked Don Klaiss, an old Oracle hand, whether his life worked now that he was running on 11i. It was almost touching to see him relax his guard. "Oh, yes," he said, and proceeded to give us example after example of how much more effective he was, now that he had his 11i dashboard.

What Larry (and, I fear, Charles) doesn't see is that finally making a 15-year-old idea work at your own company doesn't necessarily give you all that much credibility. It's interesting, but it isn't leadership.

In the interests of full disclosure, however, I must confess that if I really thought that single global instances were a great idea and a big deal, I probably would give Larry and Charles (whom I respect immensely) the benefit of the doubt. But I think that there's something specifically wrong with the idea that Oracle leads with. At the risk of trying your patience, I'll spend the next section trying to explain what that something is.

The Universal Library

About a year ago, I spent a lot of Short Takes space explaining why the idea of a single global storehouses of transaction information was far less compelling than it originally seemed. The explanation was complicated, used a lot of words like "semantics" and "representation," and would have made anybody who has read Wittgenstein carefully extremely happy.

In retrospect, I think it was pretty silly to bring out all that intellectual firepower. Because the reason why vast storehouses of corporate information aren't by themselves a good thing to spend lots of money on is completely obvious to all of us.

All of us, you see, already know what happens when you collect huge amounts of information into one place, hoping that simply by collecting the information, you'll change people's lives. You get what are called "libraries" if the information is in book form, or "the Internet," if it is in digital form.

And we all know why either of those things is not something that makes a huge, material difference to our lives. Yes, there is information there, and that's a good thing. Unfortunately, the good information is buried in huge amounts of information that is irrelevant, inaccurate, silly, out of date, tendentious, wrong, irrelevant, beside the point, and irrelevant.

The reductio ad absurdum of this experience is in the old Borges story, "The Library of Babel." In that story, you may remember, Borges imagines a library whose books contain every possible sequence of letters. All knowledge would be contained, but so what? No one would ever be able to tell you where the knowledge is.

In Borges' time, his "universal library" would actually take up more matter than existed in the then-known universe. These days, though, you might be able to put it on a grid.

Among the more pedestrian points that Borges, a librarian, was making is that a real library is much more than a storehouse. To be functional at all, it must be a place where information is constantly culled, sorted, classified, discarded, absorbed, and guided into the hands of the person who wanted it.

So, too, with any library of corporate information. The storehouse idea that SAP went to market with 15 years ago isn't adequate; it's not even clear that it's desirable. What you need is information that is organized, cleansed, culled, sorted, validated, updated, and managed. In a corporate world where almost every denizen already has too much information, you don't solve the problem of information overload by pitching all the information into one grid-based storehouse. You solve it by designing an entire system of information that can consistently bring trustworthy and relevant information to the right people at exactly the right time in a form that is meaningful to them.

The transaction systems that have been around for the past 15 years--even those that can exist on single global instances--are emphatically not systems designed to bring trustworthy and relevant, etc., etc. The only thing they do (Oracle Apps, SAP, PeopleSoft, whatever) is collect information and store it somewhere. As Laurens van der Tang, former President of Baan, used to say: "ERP systems eat information like elephants and [excrete] information like a mouse."

Bottom line, then. If Oracle wants to lead, it has to offer something that solves the problems that everybody faces, and that something is much more than a standard transaction system embedded in a single global instance.

Why the Wrong Vision Matters

When your vision is played out or insufficiently realized, the organization guided by that vision will have weaknesses. As I'll point out in the next piece on Oracle, the current Oracle product already suffers from its owners' excessive concern with creating a data model that will support a single global instance.

More relevant these days, a weak vision also weakens your plans.

Oracle has now acquired two of its three major competitors. Its plans for those competitors seem to me to be weakened by this limited vision of what transaction systems need to do.

Each of the products it is acquiring operate on quite different design principles. Each product is intended to serve different sorts of businesses and organizations, and each is designed to deliver different sorts of value. It would be entirely possible for an acquiring company to respect the design intentions of the products and make a virtue of the differences among the products. You could say that you, Oracle, actually offer consumers a choice, the kind of choice that allows consumers to maximize the benefit that they want.

But it is clear that this is not Oracle's intention at all. At the teleconference following the acquisition, Larry said that he would continue to support the products through the next version (though he got the version numbers for JD Edwards wrong). But he also said that he wanted to make a "superset" product available in three years.

Now this is basically a silly idea, like building a skyscraper on top of an art museum. (Oops, somebody did that.) It's silly because when you are building any transaction system, you've got to make decisions about how it's going to handle the basic operations in the system. These decisions are inherently exclusive: you can only have one integration strategy, one user interface, one development environment. A superset product, no matter which product is used as its base (and we can guess which one that will be), will necessarily force 2/3 of the customers to abandon their old integration strategy, user interface, etc.

That would be OK if there were a benefit to doing so. But there isn't. It would be a dead loss for the customers.

PeopleSoft and JD Edwards already ran into this problem. The original intention, surely, was to move to a superset. But when they started looking at the details, they decided that there was no point.

So why would a smart person like Larry have such a silly idea? Well, when you think that the only thing that is crucial in a transaction system is its data model, it is easy to believe in superset applications. Superset data models always make sense; just add more data elements.

And what does this have to do with leadership? Well, anyone who lives and breathes how transaction systems are used in business already knows that you don't build superset applications. By promising a superset in an impossibly short period of time, he simply tells the people who know that he isn't thinking about the problems that concern them.

Business Leadership

What problems are these? They are the problems that people run into when they want to improve business operations and they think that a transaction system would help. They are the innovations in the way of doing business that businesses want technology companies to provide.

Now, in my experience, no transaction company, even SAP, has really provided good leadership in this area. Transaction companies are good at going to companies and saying, "Teach us about your business." and they are good at writing Power Point slides that claim that problems are solved.

But, as I've said in I don't know how many Short Takes, they are not good at understanding how their brilliant innovations will actually result in benefit to their customer. There is a software value chain, and most application companies don't understand what happens after their link in the chain.

But, of a bad lot, SAP is clearly the best. Particularly with very large businesses, they seem to have the best grasp of those businesses' needs, the most comprehensive ideas about how technology can help satisfy those needs, etc. They may be no more conversant with the inner life of their interlocutor than the British diplomat was in the days of the Raj who had picked up a few phrases of Hindi. But that still makes them a better partner than the typical diplomat of that era, who "spoke Hindi" by speaking English loudly.

As long as Oracle continues to think that its fundamental job is to build large data repositories, that smidgen of business leadership is all that SAP will ever have to show.

Can Oracle Lead?

So, is it all over but the shouting in the applications area? If history is a guide, as I've said, the chances are that it is.

But it doesn't have to be. And Oracle does have some glimmerings of an idea about how to one-up SAP with a vision that is more substantial and less outmoded than the vision of a single global instance.

Remember what the basic reason for the acquisition of PeopleSoft and JD Edwards actually is. Oracle foresees a world where (grid) databases, application servers, communications, and applications are so deeply intertwined and interdependent that companies will treat them as a single stack. Then, the world of corporate information providers will be like the world of Orwell's 1984: three countries taking up the known world, forever fighting.

In their own stack, Oracle is already breaking down the barriers between database, application server, and application. The "material views" that make reporting in 11i much better than in competing products are created by exploiting a capability of the (grid) database. The views don't work on other databases. Oracle's data hubs actually work by splitting the data model out from the e-business suite and embedding it in a new database. Most of the RFID capability is built into the application server. The applications workflow engine is gradually being supplanted by workflow provided by Oracle's BPEL (Business Process Language).

If Oracle were to continue to break down the (somewhat) artificial barriers between application, application server, and database, they might have something new, which is genuinely differentiated and attractive. Combine that with a (very good) data model and the functionality that they can grab from PeopleSoft and JD Edwards, and you might have something that would be worth going to market with.

Add two more things, and I think they might be there. First, they need to recognize that most companies want to work with people who have a genuine grasp of their business problems, and they need to act on that recognition by spending money to find and develop people who have that grasp. And second, they need to take some responsibility for delivering business benefit, that is, for making sure that the value arrives at the end of the chain.

With the current vision and the current leadership, neither seems likely. But if they were to do all these things well, the applications space might have a Number Two that is worth listening to.

To see other recent Short Takes, click here for a listing.

Where is this product going?