Squabbling about who was the first to come up with “the next big thing” is a favourite game in Silicon Valley, and it is being played with gusto over the real-time enterprise. Vivek Ranadivé, the chief executive of Tibco, a Silicon Valley software firm, is miffed that Mr Khosla at Kleiner Perkins Caufield & Byers, the venture-capital firm, has appropriated his thoughts about the subject. He grumbles that the money man must have picked up the idea when Mr Khosla presented him with a leadership award a year ago.
In technology, especially, good ideas tend to pop up in several places when the time is right. But Mr Ranadivé was definitely one of the first to realise that there will be a growing need for integration: linking together disparate computer systems so information can flow more or less instantaneously. He also saw that this has to be done in a way that allows companies to change their business processes quickly.
To become real-time, companies must have an overarching “spreadsheet” that connects everything they do. But they also need good tools so they can easily change “macros”. And because users cannot keep track of all those rows and columns, they need an intuitive overview of the information they really care about.
Indian-born Mr Ranadivé was one of the first to think about this because he had seen the opposite in action—on a trading floor at Goldman Sachs, an investment bank, in the mid-1980s. The room looked like the warehouse of a disorganised computer store, Mr Ranadivé writes in his book “The Power of Now” (McGraw-Hill, 1999). In a business where every second counts, each desk was packed with as many as 18 separate monitors and several keyboards so that the trader could draw information from a range of 25 different, technologically incompatible news sources. He then had to aggregate all the information himself, often using nothing more high-tech than a pen and paper.
Let’s Take the Bus
The obvious answer to this IT chaos—to build one huge, integrated system—was impracticable, because it would have been impossible to get all those involved to agree to a common set of standards. Building a system for each trader would have been too costly and complex. Mr Ranadivé’s solution was what he calls “the information bus”—a way of reducing the data formats of the different news sources to a common denominator so that they could be displayed together.
This universal data conduit made it possible for traders to get exactly the information they wanted. The software does not send news items or stock quotes to individual traders, but “publishes” them to electronic addresses representing a subject, for example news.microsoft.nasdaq or equities.ibm.nyse. Traders can subscribe to these feeds and be updated automatically.
This way of distributing information, called “publish and subscribe”, became the basis for many digital trading systems on Wall Street. Today, the trading floor at Goldman Sachs looks very different. Only two monitors sit on a trader’s desk—one for the firm’s internal and PC applications and one for MarketSheet, a service that brings together all the information a trader needs on one screen: stock quotes, news feeds, earnings information, graphs.
Mr Ranadivé’s first start-up, Teknekron Software Systems, sold such messaging software to the financial-services sector. In 1997, he founded Tibco (The Information Bus Company), because he was sure that firms other than those on Wall Street would soon need similar technology. For one thing, capital markets, where most transactions are time-sensitive, have traditionally adopted new technology before the rest of the economy does. But the need to tie together disparate computer systems is even more noticeable outside the financial world. Companies have accumulated layers and layers of IT systems, many of which are unable to talk to each other.
Take databases. It is not at all unusual, even in high-tech firms, for information about a single customer to be spread across a number of different and incompatible databases. Until recently, Oracle, a software giant that makes database programs, operated more than 100 different databases worldwide for its own customers and 70 for its human-resources department. If anyone wanted to find out the exact number of Oracle employees, it would take weeks of searching—and by the time the answer was found, it would already be out of date.
Customers of a mobile-phone company can easily discover how well-integrated the company is. All they have to do is telephone their mobile-phone operator and ask how much airtime they have used this month. The chances are that the people on the customer-service desk will be unable to tell them, because their computer is not linked to the firm’s main database. When a customer has ordered a PC and it does not arrive within a week, as promised, the person on the other end of the line will probably be unaware that this is because the company is short of a single part, and that the caller could have his machine within days if he changed his order just slightly.
One explanation for this digital Tower of Babel is that information technology is a complex beast. Bridges between applications usually have to be hand-coded—a tedious and unpopular job—by in-house programmers or systems integrators. But it is not just technology that is to blame; political and economic reasons are at least as important, argues Ole Hanseth, a scientist at the University of Oslo who has studied the dynamics of IT infrastructure in global organisations. Departments have built their own systems and are wary of integrating them with others because they fear they will lose some of their autonomy and thus their power.
Financial incentives play a big role too, according to Gartner’s Michael Maoz. They can turn integration into a common good that nobody wants to provide. American consumers often wonder why, if they ring their telephone company and are handed from one call centre to another, they always have to repeat their contact information and describe the problem all over again. This happens mainly for accounting reasons: if the first call centre were to hand over all the information to the second one, it would bear the main cost of the call and make the second look more efficient.
These sorts of organisational problems also help to explain why the first serious effort to merge “data silos” was often painful. The effort in question was to implement enterprise resource planning (ERP) systems—complex software suites of interlinked applications that were pioneered by SAP, a German software giant. The idea was to use a unified database to make sure that different programs, say for financial planning or human resources, worked with the same information. In fact, as Peter Zencke, an SAP board member, points out, the “R” in R/3, the name of his company’s ERP system, stands for “real-time”.
Yet ERP systems cleaned up only part of the IT mess, and created new problems of their own, mainly because they are inherently inflexible. They have their own way of managing the information in databases which cannot be easily changed or shared with other applications. In effect, explains John Hagel, a noted American e-commerce expert, ERP systems have replaced “fragmented unit silos with more integrated, but nonetheless restrictive enterprise silos”. Many companies therefore get the worst of both worlds.
Meanwhile, ever-increasing competition and the recession have been pushing firms to become both more integrated and more flexible. If they want a global organisation to act as one, or to show a single face to their customers, they have to streamline their information flow. And given rapidly changing markets and the endless succession of acquisitions, divestitures and new strategic partnerships, they need to be able to change their business processes swiftly.
The technology provided by Tibco and similar enterprise application integration (EAI) companies such as CrossWorlds (recently bought by IBM), SeeBeyond and Vitria goes a long way towards resolving the tension between integration and flexibility. Conceptually, their products are all similar to Mr Ranadivé’s “information bus”, except that they do not connect news sources with traders but machines with other machines.
But the programs are not just dumb conduits. They let firms embed business processes into the information flow; for example, how an order is processed or when a supplier is paid. And this integration software now comes with graphical tools, so users do not need to be seasoned hackers to program and change business processes, as they had to be with ERP systems.
Web Services to the Fore
Yet EAI and related software, too, has its drawbacks. It is expensive, and so far it connects only a limited set of applications. But its biggest disadvantage, according to Don Ferguson, an IBM research fellow, is that it does not provide the flexibility of changing business processes in real time. This is where web services come in. This technology, a set of standards based on XML (Extensible Markup Language, used for web pages), allows firms to integrate their systems even if they do not have their own networks and lack the money to buy expensive EAI software. Web services try to standardise integration by turning the Internet into a universal data bus. Programs publish their data feeds following a set of Internet standards so that other programs can easily subscribe to them.
The central piece of code in the real-time enterprise is likely to be a program called an application server—an operating system of sorts for web services, just as Windows is the operating system for PC applications. Unsurprisingly, IBM offers such a product, called WebSphere, of which Mr Ferguson is the architect. But there are other vendors, such as BEA, iPlanet and Oracle. Even Microsoft’s .NET framework belongs in this category, although it is based on a different programming language.
The big hope of web services is that they will make the integration of computer systems across firms much easier, says Charles Fitzgerald, a leading software strategist at Microsoft. A favourite example is that it took Southwest Airlines and Dollar Rent A Car only a month to hook up their systems so that visitors to Southwest’s website could easily reserve a rented car after they had bought a plane ticket. With traditional EAI technology, the same feat would have taken months and cost millions of dollars.
IBM’s Mr Ferguson goes even further. Before too long, he says, web services will be able to team up as required with the help of software agents to solve a one-off problem such as making the necessary arrangements if a flight gets cancelled. A collection of web services would take the place of a travel agent using the phone or the fax: try to get another flight, book new ground transport, perhaps reserve a hotel room.
Yet some software firms doubt that this hyper-flexible world of web services will come to pass any time soon. In fact, they are convinced that companies will always need a helping hand to make their disparate systems work together. To avoid complications, buy everything—or at least the core applications—from us, says Oracle. More interestingly, Asera has positioned itself as a kind of super-integrator. Firms can install its “eBusiness Operating System”, a collection of the available integration technologies, and hire its consultants to speed up their transformation into a real-time enterprise.
There are also lots of start-ups with their own angle on real-time computing. KnowNow software allows users to build publish-and-subscribe services delivered over the Internet. Bang Networks has developed technology to update information on web pages continuously. Kenamea’s products, among other things, speed up web-based applications.
Horses for Courses
But how will people work with all these integrated systems? The “user interface” of the real-time enterprise will come in different shapes and sizes, depending on the role of the employee and the device concerned. For service technicians it might be nothing more than a display on a pager that tells them where to drive next. For office workers it will probably be an “enterprise portal”, giving personalised access to corporate information and applications. And senior managers will have their “digital dashboards”, showing them in real time how their business is doing.
Software experts are already speculating about the way corporate IT systems will look in 20 years’ time, when the “legacy” systems (the term used for old applications) will, with luck, have disappeared and EAI vendors will be history. In all probability, corporate IT will by then have become amazingly lightweight. Companies won’t need much more than a “real-time manager”, to use a term coined by Thomas Berquist, an analyst with Goldman Sachs. Everything else could be outsourced via web services, except perhaps a few core applications. And firms will be able to tap into many more data sources in future.
Copyright © 2002 The Economist Newspaper and The Economist Group. All rights reserved.