Blinded By the Light

In the run up to JavaOne this year, most of the emails we received focused on the new wave of Ajax tooling for Java. Given the build up around Web 2.0, we’re not terribly surprised that rich clients have become the dominant theme.

In so doing, the Java community has gone full circle. At our first JavaOne, nearly a decade ago, most of the attention centered on the possibilities of Java applets for animating web pages. Yet, at that same event, we saw a startup called WebLogic demonstrate how Java on the server could provide the solution for powering web sites that could host thousands of concurrent visitors interacting with highly complex enterprise applications.

As it turned out, rich Internet clients in 1997 proved an idea before its time for two reasons: bandwidth was lacking, and religious wars were being fought between Sun and Microsoft for the client. By contrast, WebLogic made the right gamble on technologies that would later become part of J2EE. The company was subsequently acquired by BEA, and the rest was history.

The renewed attention of Java on the client can be summed up in two words: Google Maps. There, the recipe for success started with a technology that was simple, and a business model that made the technology accessible. But the clincher wasn’t the technology alone, but the fact that mashing up Google Maps with other apps satisfied an unmet market need for the localized Internet. Just as Zagats proved that diners were hungry for restaurant guides reflecting the opinions of actual customers, Google Maps has proven that there is a real appetite for the localized Internet. For instance, people or advertisers are more than willing to pay for information regarding how much homes are selling for in their neighborhoods.

So we’re not surprised that this year, the Java community is fixated once more on the client. Given that Ajax, the poster child technology for Web 2.0 is platform agnostic, it provides a fat and fashionable bandwagon for Java providers to hop aboard. And on this go round, there’s plenty of bandwidth. And, aside from rivalry of NetBeans and Eclipse factions in the Java world, religious wars on the client are for the most part artifacts of the past.

Nonetheless, hype is at best a trailing indicator of market need. Consequently, while the underlying theme of JavaOne 2006 is Web 2.0, we learned that ParaSoft, a provider of web services testing, was fielding more queries at this conference than in past years. Current attention on the client notwithstanding, web services and service-oriented architectures (SOAs) present the possibility for opening up new business possibilities through integration, just as Google Maps did with the rich client.

As you lower the barriers, more people come in, and there are more chances for things to go wrong. And so there are problems to solve when dealing SOAs. Believe it or not, that’s a sign of progress.

Nonetheless, it underscores our prediction that next year at JavaOne, our attention will probably drift back to connectivity, just as we previously did with J2EE when it was young. What goes around comes around.

The Enterprise Mashup

Last week, SAP and Microsoft proved what was possible when you get a couple of 16-ton gorillas to march in sync. They introduced what’s for now, the enterprise’s ultimate mashup: ERP and Microsoft Office. And today, SAP is making similar announcements with IBM.

Duet, the official name of the joint SAP/Microsoft offering, is designed to push SAP’s installed base past the core 10 – 20% of the enterprise. The value proposition is, wouldn’t it be great if a manager who needs budget rollups once or twice a week gets them fed to the Excel spreadsheets that is his/her de facto standard? Or what about the road warrior who would rather stay in Outlook than fiddle with the CRM system?

Clearly, people are more productive when they work with the software they know, and don’t have to go elsewhere to cut and paste data or rely on a custom interface. And the idea of an all-in-one desktop is nothing new. It drove emergence of the first word processor/spreadsheet/calendar bundles 20 years ago, and got a shot in the arm once the Mac GUI enabled you to cut and paste data from one app to another.

But finding a way to mass-produce enterprise mashups encompassing business functionality has remained elusive.

The emergence of services-oriented architectures, which supported loosely coupled and dynamic connections, moved the integration ball off dead center. SOAs enabled SAP and Microsoft to do Duet, and for IBM to knit SAP into IBM Lotus Workplace and WebSphere Portal.

While SOA makes integration easier, it’s still not easy. Duet, which is actually more of a conventional packaged product with specific functionality that you install and configure, required two years of development for its first modest HR-focused release. By contrast, IBM, which took a toolkit approach, got the offering out in only six months, but it leaves actual integration to the customer.

Either way, you can’t just pick the pieces you need on the fly; either the vendor must prepackage them or you must integrate them yourself.

In large part this explains the popularity of Ajax-style mashups. That’s Zimbra’s approach to groupware, which seeks to unify two of the most popular enterprise apps: email and web browsing. Because Ajax focuses on rich UI rather than transaction integration, and because it uses some of the best-known web languages around, it’s a pretty fast way to make enterprise mashups.

The drawbacks of course are that Ajax cannot integrate at process level, and for now, programming standards are nonexistent. That makes Ajax a great tactical approach if you want to blend a couple basic functions quickly, like enhancing a customer segmentation analysis with a geographic system to literally map where your market is concentrated. But Ajax’s loosey-goosey nature for now renders it difficult if not impossible to replicate and scale at enterprise level.

Hope may be on the way. Future efforts to componentize web services might reduce lead time for SOA-based unified desktops by making services more plug and play. Meanwhile BPM (business process management) approaches might help streamline mashups at the process level. Meanwhile, Open Ajax will have its first summit next week to hash out a roadmap for harmonizing Ajax tooling frameworks.

Enterprise mashups have a long way to go before we can simply plug and play them together.