Today, the use of rich Internet application (RIA) technologies is planting the seeds for a return to client/server architectures where large clients are installed on the desktop.
This convergence of solutions is called the Fit Client or Desktop 2.0. While RIA technology is the darling of the Internet today, it’s likely that client/server architectures based on RIA technologies will become the dominant solution of tomorrow.
Examples of Fit Client platforms include Adobe AIR, Google Gears, Curl Nitro, and Mozilla Prism -- a list that is bound to expand over the next year. These solutions re-envision RIA as a new kind of client/server technology that spreads the processing load between the client and the server and offers end users a richer experience and more responsiveness.
Since the introduction of the Internet and the World Wide Web around 1994, the old client/server architectures, typified by large Visual Basic or PowerBuilder software clients directly accessing a shared database, have fallen by the wayside.
These first client/server platforms were attractive because they provided a richer experience for end-users by moving the software to the client machine, instead of requiring roundtrip processing of all user input at the server. If this sounds familiar it's probably because this is exactly the argument used to promote rich Internet application technologies, such as Adobe Flex, Ajax, Microsoft Silverlight, and Curl.
The primary difference between the new client/server architecture of today and those of yesterday is threefold:
- Client applications are no longer accessing databases directly. Instead they access data services embodied in the ethos of Web 2.0.
- Client/server applications do not run directly on the operating systems, but on an intermediary platform called a runtime. This makes portability easier -- a lesson learned in large part from Java.
- With broadband, it’s no longer an issue to download an application that is a couple of MB or more. Distribution technologies even allow in-place updates of software right at the client machines, so distribution has become less of an issue than it was 10 years ago.
The Web with HTML is not going to go away. Neither are pure RIA solutions that run inside a browser and are downloaded each time they are used. Rather, all of these solutions will exist in parallel, offering organizations a spectrum of deployment and interaction options.
We didn't give up walking when the bike was invented; or biking when the automobile was invented; or driving when the airplane was invented. All of these travel modes exist at the same time and are used appropriately. The same will be true for network computing. We won’t abandon the Web for RIA, or RIA for the new client/server architectures. We’ll use them all appropriately.
— Richard Monson-Haefel, VP of Developer Relations, Curl Inc.