The environments to which developers must deploy their wares are becoming increasingly more complex, despite a general trend towards consolidation of both platforms and vendors. As a result, the costs of developing, maintaining, and managing those environments may end up offsetting any savings realised by the Web service promise to reduce integration costs.
On the mobile client side, between the lack of presentation layer standards for wireless devices and end users whose mobile habits and acquisitions have evolved outside of the watchful eyes of IT departments, corporate developers now face a daunting number of devices, presentation formats, and end user expectations.
On the server side, the old days of simple terminal-host computing gave way to two-tier client/server architectures, then three-tier client/server architectures, and now, a thick middleware layer with boundaries so gray, it's hard to distinguish between the middleware, the applications, the operating system, the clients, servers, and any other software that might orbit the application stack's universe.
Normally, developers and IT shops must, on their own, research the various nuances of their environments that lead to instability or unpredictability in their application stack. Maintaining that research in a dynamic environment with lots of moving parts can be an incredibly resource intensive effort.
This complexity has created an opportunity for a handful of small solution providers to study those complexities and package the resulting knowledge into software designed to not only eliminate all that work, but also resolve the issues that arise from those complexities.
Embodying knowledge in software in a way that automates that knowledge isn't a new idea. CASE tools, code generators, business intelligence solutions, and accounting packages automate the application of well-known methodologies to the problems addressed by each of those disciplines. But when it comes to managing the plethora of interdependencies that can exist anywhere in the application stack, the potential problems seem so environment-specific that few companies have taken on the challenge of trying to ameliorate them. That's changing.
At Gartner Symposium/ITxpo 2002, I spotted two such applications.
Air2Web's Mobile Internet Platform
One is from a company called Air2Web. On the back of the T-shirts that Air2Web handed out at the Symposium/ITxpo is the following: "65 million digital cell phone users. 4 million WAP phone enthusiasts. 150,000 Palm fanatics. 50,000 Rim page loyalists. 6,000 operating companies. 200 carriers. 11 networks. 1 application." The T-Shirt is alluding to the next generation of an old problem familiar to commercial Web developers. If you look at the source code of just about any Web page, you will see that the code includes if-then-else statements that are, at the very least, specific to Netscape and Internet Explorer. Sometimes, they go even further to deal with different versions of each.
The next wave of dependencies is around the corner. As the back of the T-Shirt indicates, there are many types of mobile devices to which you might have to deploy your applications. Now that you've developed and are maintaining two sets of code for Netscape and IE, are you ready to develop a whole bunch more to deal with the presentation specifics of WAP, Palm's Web clipping, BlackBerry, and J2ME? What about dealing with each of those platforms' device or network dependencies? For example, J2ME may be a write once, run anywhere platform. But unlike its big sister J2SE, which has SWING (a built-in API for the user interface), developers are on their own for J2ME. As a result, a J2ME application with a user interface may not look or work the same on two different J2ME-based phones.
Air2Web claims to have solved this problem by studying the nuances of the client platforms, the devices they run on, and networks they attach to in a way that allows developers who target those platforms to develop their applications just once and with virtually no regard for what the target platforms are.
LoudCloud to OpsWare
Speaking of Netscape, another product I spotted at Symposium/ITxpo is the brainchild of Marc Andreesen. Somewhere in Andreesen's travels, he must have seen companies, hosting facilities, and consultancies wrestling with similar problems within the application stack. He also saw that those companies reinventing the wheel--taking precious time to understand those interdependencies. To a large extent, the company he founded---LoudCloud---was based on the notion that those interdependencies could be learned, captured, and automated in a way that greatly reduces the cost of systems management. Having such a technology could have resulted in LoudCloud running circles around competing managed hosting services.
But, according to company officials, LoudCloud had difficulty getting traction in a crowded market. Andreesen took the intellectual property that was meant to propel LoudCloud to greatness, and bottled it into an application, OpsWare, that is available to anyone who's tired of researching the interdependencies between operating systems, applications servers, database servers and content management systems.
We've seen this class of application before. But, I don't remember seeing it in the context of automating interoperability and compatibility in constantly changing software environments. As I see more of these solutions---especially the ones that hold promise for dramatically reducing costs or taking that dreaded grunt work out of keeping applications running smoothly --- I'll be sure to report on them.
What grunt work would you like to see eliminated from your IT load? Or, have you found tools like the ones I've found that are saving you enormous amounts of time and money? Share your thoughts and ideas with your fellow readers by using TalkBack below, or write to edit@zdnet.com.au.
13%
1%
