Columns

Warm and fuzzies

Anybody who recalls the old Holiday Inn ad campaign that "The best surprise is no surprise" reveals two things: One, they are probably older than the typical Java developer today; and two, they would rather not have to sweat unnecessary details -- especially when building large systems.

"Commodity" is a very charged word to IT vendors. Mention it around the likes of Oracle or SAP, and you conjure up visions of dwindling sales and profits due to products that have either saturated their primary markets or lost their uniqueness.

Mention the term commodity to a CIO or development team manager, and the first things that come to mind are "no surprise," followed by "investment protection." In commodity markets, users gain the upper hand because they have a wide choice of similar products. Not only are prices better, but users start to realize that their IT destinies are no longer tied to a single vendor. You might have a "buggy" version, lousy service or your vendor goes bust -- but don't worry, the investment is still safe. Traditionally, that was the case with desktop PCs. Few gave a thought to replacing Compaqs with Dells, or adding a few IBM Thinkpads to the mix.

>Maybe you had to re-install software -- some- thing that could be automated -- but at least you could keep what you had. The same has become pretty much true in servers. Swap an HP-UX server with Sun/Solaris and your Unix gurus might have to recompile programs; but as long as it's a major brand of Unix (or soon, Linux), you should be able to keep what you have running.

Although things aren't quite as transparent on the database side (R/3 still doesn't run on Sybase, and probably never will), ongoing consolidation toward IBM, Oracle and Microsoft may render the question moot. Switch to a different relational database, and the flavors of SQL might be slightly different; but, in most cases, you won't have to replace the core systems.

The common thread is that each of these investments is a building block or, to use another charged word, a "utility." These products provide basic services that allow development organizations to build the stuff that really adds value: e-commerce systems, enterprise applications or workgroup business systems.

Logically, Web app servers should fit that definition well. They are a means to an end for distributing or deploying value-added business functionality to users via the Web. So why should they be a destination buy?

In the early days, it was because the products were so different. Products that came to life before Java, such as Allaire ColdFusion, Bluestone's Sapphire/Web, Netscape/ Kiva and NetDynamics, each had their own proprietary development environment. Others, such as SilverStream, embraced Java early, but with their own unique Java classes and means of generating HTML pages. Meanwhile, WebLogic prided itself on the fact that it embraced Enterprise JavaBeans (EJBs) before there was a real spec. The result was that if you developed app server logic in any of these products, the code -- or in the case of Java, the classes -- was proprietary.

As the Web turns, with e-commerce systems emerging not as extensions of transaction systems but as core engines of enterprise business, customers expect Web application servers to take on enterprise-like features. On the short list are features like transaction protection; clustering, failover and load balancing; messaging support; and APIs to enterprise resource planning (ERP) systems, databases and other information sources. The result is less differentiation in bells and whistles. And, although many app servers still carry their own IDEs, many allow you to plug in the Java IDE of your choice.

Sun's recently released Java 2 Enterprise Edition (J2EE) threatens to further flatten the playing field, at least for distributed Java support. Instead, the choice of an app server will likely lodge on the more basic issue of which distributed architecture vision to embrace: Java, Distributed COM+ or CORBA. Web app servers, like databases, remain critical building blocks, but they are becoming more and more of a commodity -- with a few caveats. You won't get plug and play, so forget about mixing one vendor's EJB server with another's Java Server Pages feature. Then again, few sane systems administrators would want such unnecessary complexity anyway. Commoditization might be a letdown for app server marketers, but it is a sign of maturity when buying Web application infrastructure tools no longer amounts to a riverboat gamble.

About the Author

Tony Baer is principal with onStrategies, a New York-based consulting firm, and editor of Computer Finance, a monthly journal on IT economics. He can be reached via e-mail at [email protected].