Network Computingwww.networkcomputing.com

The Internet -- more specifically, the Web -- has changed the equation. Back-office packages are acquiring Web interfaces, and in the process, IT departments are gaining more flexible systems. But how does that flexibility affect the overall cost of ownership or the ability to deploy custom applications on top of base packages? It's safe to say that while most IT departments appreciate the ability to use a Web interface to these packages, most haven't given up the fat-client-side user interface yet.

Microsoft Corp., with Exchange 2000, and Lotus Development Corp., with Domino, are both working to Web-enable their systems and to simplify development. Microsoft takes this a step further with its Tahoe project. Tahoe combines portal technology with document-management and search-engine capabilities. With this product, customers can organize data on the Web straight out of the box, or the system can be customized and extended as needed.

Notwithstanding the efforts of vendors to shift their back-office applications toward the Web, there are good reasons for not dumping the client-side applications. First, a good deal of user training is invested in those apps, so swapping them out will mean retraining users. Then there's the Web interface. JavaScript and other such browser improvements have led to better user interfaces, but there's still some ground to make up before capabilities are on a par with existing client applications.

It's not just the user interface of back-office applications that's changing. At the back end, these applications are being retrofitted to use external directory and authentication systems. The days of every application requiring the use of its own user-management environment are numbered.

Custom App Revolution

While back-office applications are evolving, it's the customized line-of-business applications that are undergoing revolutionary change. As always, the hype in our business outstrips the reality of the revolution, but the case for deploying custom applications definitely has been bolstered. Two factors in particular make the custom application more attractive. First, the pervasiveness of the Web means that applications can be extended to reach suppliers and customers. The potential to improve services to, and relationships with, both groups is alluring. And it's the main reason you're seeing tons of coverage of CRM (customer relationship management) software right now. A recent Meta Group survey of 300 companies showed that 75 percent plan to increase their investments in this area in 2001. Eighty percent of the companies said they'd already had some success with their CRM efforts. At the same time, many of the companies surveyed admitted that they have not yet fully integrated their CRM programs into their overall e-commerce efforts, which include Web storefronts and self-service.

Second, the Web environment is more amenable to custom development than were the client/server environments of the past. Smoother development and the potential to reach customers and suppliers more easily make the case for deploying custom applications very tempting. But for those who have deployed client/server applications using non-Web-related development systems, the issues at hand are more complex. The Holy Grail of the Web-enabled Information Age starts with a few centralized, coherent repositories of data. These databases are tapped for information on sales, production, supplies, customer relations and much more. In theory, management has at its fingertips all the data needed to make informed decisions on these issues and others. Customers and employees throughout the company also have access to exactly the data they need (no more and no less), ultimately spurring the company's growth. The point is simply that if you've implemented systems the old way, there's a good chance you'll eventually scrap nearly everything you've done up to now.

OK, that's a somewhat extreme gloom-and-doom view, but at a minimum, deployed applications' days may be numbered. For organizations that have chosen well, existing databases can still be used as the basis for new applications. A large part of the industry is devoted to helping organizations make the transition. The upside is that the manageability gains found in a Web-centric IT shop can be significant enough to offset at least some of the pain of redeploying client/server apps. For example, the Web model inherently provides more centralized control over application deployment, and is not as sensitive to network performance variations as are most client/server environments.

To a great degree, the simplifications of the Web-centric model are found mostly outside the data center. Client/server has become a "bad" term, but clients and servers still exist; so does middleware, to tie application-specific servers to general-purpose databases. Java is the development language of choice, and everyone from IBM, Oracle and Sun down to a slew of start-ups offers tools and services to develop custom applications or customize prefab applications to fit customers' needs. In a sense, that's the real revolution. We've all finally agreed on a development system and a delivery vehicle.

But while agreeing on a development environment is good, it isn't necessarily sufficient for success. Compared with C++ and certainly with COBOL, Java is still wet behind the ears. Luckily, the Web mania of the past few years has forced Java to grow up very quickly. It has also resulted in rapidly maturing browser technologies, such as DHTML, XML/XSL and JavaScript. Enterprise JavaBeans provides the middleware layer needed for application development, and Java's security model and overall scope have been reinforced to enterprise strength. However, some pieces are still missing.

Directories capable of universally matching users to resources and a widely deployed PKI (public key infrastructure) are both works in progress. Internally, most enterprises have quite a few directories in place, so even if you haven't developed client/ server applications, you may have to reconcile a fair amount of infrastructure to deliver Web-enabled applications throughout your organization.

Companies To Watch

Just as in the old client/server days, there's a choice to be made in buying applications. There's also the option of renting apps as a service, but that's a different story. For smaller shops, more powerful canned applications are the hot ticket to providing customers with more services, as well as to better tracking of internal resources.

Midtier companies may have the most interesting choices to make. Traditionally, large ERP (enterprise resource planning) systems have simply been too big, too costly and too time-consuming for midsize companies to implement. A considerable number of vendors are now using the Web to craft offerings for this segment of the market. While it's nice to have choices, many of these vendors are start-ups and, amid the current market conditions, jumping into bed with these vendors for critical business software is a scary proposition. But solid vendors are offering excellent Web-based applications that can be tailored to each company's needs. These applications often start with an e-commerce initiative that gets the company onto the Web one way or another. Continuing down the e-commerce path is generally determined by customer demand. Perhaps the biggest trick here is developing a cohesive system as applications are added. And, as has often been the case for midsize companies, Microsoft may hold the solution (see "It's Microsoft -- No Java Here").

What About ERP?

For large enterprises, ERP vendors are still the best game in town, but plenty of Web-application vendors also are offering tempting systems. Companies such as Oracle, PeopleSoft and SAP, as well as a horde of start-ups, bring prebuilt modules to the table, and then perform some customization to meet the needs of each enterprise. Consultancies still do much of this work, but increasingly the vendors themselves will sell professional services along with their applications. There's a ton of money to be made in, and to spend on, professional services, and the app vendors have much stronger relationships with end customers. Within this group, the traditional ERP vendors are working to add Internet functionality to their software suites. The Web-application vendors, meanwhile, have created front-office Web apps, and many have had notable sales success with them. According to a study conducted by Advanced Marketing Research, the ERP market -- expected to top $20 billion in revenue in 2000 -- will garner more of its money from newer strategic functions such as EBR (electronic batch record) systems, SCM (supply-chain management) and e-commerce over the next few years. Currently, EBR systems, SCM and e-commerce account for less than 5 percent of the market's total revenue. But by 2004, that trio will total 15 percent of the $21.4 billion market, while approximately 55 percent will be derived from traditional ERP applications.

The big three ERP vendors -- Oracle, PeopleSoft and SAP -- have all released Internet versions of their software suites this year. These new versions feature a number of e-commerce extensions; some are the result of acquisitions--for example, PeopleSoft's latest release includes CRM applications it obtained by purchasing Vantive Corp. last year -- while others were developed in-house. Because the ERP vendors were somewhat late getting into the Web game, they've exposed APIs over the past few years. Web-application vendors have used these APIs to tie their front-office applications back into existing ERP systems.

While it should be possible to build a cohesive system that includes both ERP functions and front-office Web applications, it isn't easy, and one-stop shopping is out of the question. However, organizations can now at least begin to head toward that goal, with the hope of one day actually attaining it. Meanwhile, everyone has work to do. The ERP systems themselves, while Web-enabled, aren't necessarily Web-friendly. Some of the Web versions are little more than the old client/server interfaces running within a browser.