Thick or Thin? One Vision of the Future of Pocket PCs

[Jason Dunn]

<a href="http://articles.pocketnow.com/content.cgi?db=articles&id=69">http://articles.pocketnow.com/content.cgi?db=articles&id=69</a><br /><br />Russ Smith from PocketNow.com has written a column on the concept of thin client computing. It's a good read. The Larry Ellison-esque concept of thin clients everywhere and an Oracle server in every home remains a losing bet. Thick clients with local processing, storage, and functionality are the future of mobile computing. Bank on it!<br /><br />"If you've been around computer technology for any length of time, you've probably heard the thick or thin debate in some form or another. It's basically the question of whether the computing power resides in the machine right in front of you or whether it resides somewhere else, usually working on behalf of a number of "workstations" at the same time. In the old days (when dinosaurs roamed the earth, or at least took up whole floors of air-conditioned space), it was a debate between "distributed processing" (processing done by the "big iron" mainframe computer and "distributed" to local "terminals;" also called "time-sharing," because everyone shared the processing time of the mainframe) and "local processing." This debate essentially ended with the widespread use of desktop PCs. The advantages of being able to do a quick 1-2-3 spreadsheet, as opposed to whatever passed for a calculation engine on the mainframe, were just too great."

[Perry Reed]

I prefer the idea of thin computing + thick "improved" computing.

In other words, the client can do most of the processing itself, but when it has access to the server, it can offload the processing to the server to do bigger things or do them faster.

For example, a PocketPC w/o server connection could do some limited voice recognition. Maybe recognize some basic commands at some reasonable speed. But, when connected (wirelessly, of course) to a server, could achieve full voice activation, including complex commands and voice dictation.

I've never thought of thin vs. thick computing as an either/or proposition.

[Jason Dunn]

I agree with you Grabaclue - when within "reach" of more resources, the client would access them - more storage, more bandwidth, etc.

[Perry Reed]

I agree with you Grabaclue - when within "reach" of more resources, the client would access them - more storage, more bandwidth, etc.


Exactly! It would basically "upgrade" itself with whatever services become available upon connection. This might also open up a whole new category of "web services", something that Microsoft is pushing heavily.

[Russ Smith]

That's basically what I had in mind when I wrote the piece. I got tired real quick with Larry arguing that thin was the only way to go, but I also didn't like the idea of having to keep adding hardware, more processing power, and all the rest, just so my handheld can keep racing to catch up with laptops. When I saw the Mira, it all clicked. Why not the best of both worlds. Thick enough to run everything you need when all you have is your hand-held along with the hardware and software you need to integrate networked resources as they're available.

[dochall]

True thin computing is a non starter for handheld anyway. The mobility of the handheld means that at some point bandwidth is going to cost and downloading complete apps that you wanted is going to be too expensive and even with 3g too slow. And of course at some point a connection is just not going to be available (anybody use their ppc on the underground?)

However just continuing the purely fat standalone mechanism doesn't cut the mustard either. Although not directly the same we are close to the situation before LANs became ubiquotous. Fair enough we can share files but we are limited to the processing power that is within the machine.

.NET really gives us a standards based method of moving into connected Fat client using web services. Having a standard based method of extending the processing power of the handheld means that we don't have to be limited by the processing power in our hand. This, of course, takes us above the traditional fat and thin client debate and into a whole new area giving the best of both worlds.

[LarDude]

In this article, there is the following statement:
"...In the old days (when dinosaurs roamed the earth, or at least took up whole floors of air-conditioned space), it was a debate between "distributed processing" (processing done by the "big iron" mainframe computer and "distributed" to local "terminals;" also called "time-sharing," because everyone shared the processing time of the mainframe) and "local processing."

Is that what "distributed processing" really means?

For example, if one looks at the M$ glossary (http://www.microsoft.com/net/defined/glossary.asp), "distributed computing" is defined as:
"Distributed computing is a programming model in which processing occurs in many different places (or nodes) around a network. Processing can occur wherever it makes the most sense, whether that is on a server, Web site, personal computer, handheld device, or other smart device. In contrast to the two-node system—the client and the centralized server—prevalent today, Microsoft .NET uses distributed computing."