James Fallows makes an interesting point in his "Another step toward the online 'cloud computing' life" post:
Web-based computing has a small disadvantage: working with an online program like, say, Writely (now Google Docs) is slower than using one based on your own machine, since info must constantly go back and forth from a remote server.
In other words, there's a certain value in client/server computing--now considered passe--since the client can perform work locally without always going back to the server. This value add has been true for years, and remains true now. At this point we've been through four generations of client/server architecture (that I know of--there may be more). The first is shrouded in arcane minicomputer history, the second is well-known, the third has been around for several years, and the fourth is happening as we speak.
The first began in the late 1970's, when Wang Laboratories was the equivalent of a dot.com wonder, growing at 100% a year and challenging the incumbent at the time, IBM, within the office market. Its success was due to the combination of a great product (word processor) and a smart architecture (client/server, although no one called it that at the time). Each Wang workstation contained a Zilog Z80 chip, which did a lot of the graphics work required when displaying and editing a document. For example, inserting a character would cause the following characters to move one position to the right, and therefore sometimes cause a word to jump to the next line. Because the Z80 did much of the work, the workstation talked to the server after only every tenth keystroke or so.
This smart design was Wang's secret weapon. For example, although DEC eventually got into the office market, all Wang had to do to beat it in a competitive situation was to suggest to the prospect that it run a test under a heavy workload. Having forty secretaries concurrently pound out memos would bring a VAX to its knees, because every keystroke was a CPU interrupt. On a Wang VS minicomputer, due to the Z80 doing much of the work, 200 typing secretaries was a typical and easily handled workload. Unlike the VAX, the Wang system didn't have a dumb terminal as a front end--it had a smart terminal.
The next smart terminal was the PC, functioning as the front end (or client) to a server--hence the name, "client/server." The heyday for this architecture was in the early to mid-1990's. However, as the Internet became popular, the dumb terminal took over again, as enterprises liked the idea that they didn't have to install programs on their PC as long as they had a browser.
However, the siren song of the client/server architecture beckoned once again, and Ajax started to become the standard way to construct Web applications. I would contend that this third generation of client/server design is starting to give way to a fourth: a Rich Internet Applications (built with Adobe AIR or Microsoft Silverlight 2.0) front end and a SaaS-based back end. I first saw this last year at the Salesforce.com's Dreamforce conference. Even though Adobe AIR was in its early days, some small Salesforce.com partners were already demonstrating solutions that used this new RIA frontend/SaaS backend architecture. I remember thinking, "They must have built this in six weeks," and then thinking, "Given that kind of coding speed, this thing has legs."
So while the market is still fixated on Ajax, start thinking about the next wave of client/server architecture--I predict RIA/SaaS will be the way many vendors and enterprises will be building applications in the years to come.
Wang VS, Z80 chips, and Silverlight - We've come full circle.
Posted by: Jim Craig | Wednesday, April 30, 2008 at 05:33 PM
I miss it and I look forward to a day when I myself no longer have a computer tied to my hind quarters right along with the cell phone. I am not condemning these things, just stating what they are, "convienences". Just like anything else.
Posted by: Buy Online Rx | Thursday, October 14, 2010 at 12:04 PM